August 2005 - Development - lists.drupal.org

[drupal-devel] [bug] Upgrade from 4.4 to 4.5.1 creates incorrect columns in menu table when using PostgreSQL
by gordon 09 Aug '05

09 Aug '05

Issue status update for http://drupal.org/node/14928 Post a follow up: http://drupal.org/project/comments/add/14928 Project: Drupal Version: 4.5.1 Component: postgresql database Category: bug reports Priority: normal Assigned to: Anonymous Reported by: Anonymous Updated by: gordon -Status: active +Status: patch (code needs review) should be marked as patch gordon Previous comments: ------------------------------------------------------------------------ Thu, 30 Dec 2004 05:30:15 +0000 : Anonymous Attachment: http://drupal.org/files/issues/updates.inc_2.patch (956 bytes) I just upgraded my installation from 4.4 to 4.5.1 and got these errors when trying to administer the image module: pg_query(): query failed: ERROR: invalid input syntax for integer They were all for the menu table. On investigation, i found this in database/updates.inc: if ($GLOBALS['db_type'] == 'mysql') { $ret[] = update_sql("ALTER TABLE {menu} ADD description varchar(255) DEFAULT '' NOT NULL"); } else { $ret[] = update_sql("ALTER TABLE {menu} ADD description smallint"); $ret[] = update_sql("ALTER TABLE {menu} ALTER COLUMN description SET DEFAULT '0'"); $ret[] = update_sql("UPDATE {menu} SET description = '0'"); $ret[] = update_sql("ALTER TABLE {menu} ALTER COLUMN description SET NOT NULL"); } WTH??? How is a smallint on other databases equivalent to varchar on mysql? Attached is a patch to fix the updates.inc file.

1 0

[drupal-devel] [bug] user_login calls user_deny with empty variable
by gordon 09 Aug '05

09 Aug '05

Issue status update for http://drupal.org/node/11031 Post a follow up: http://drupal.org/project/comments/add/11031 Project: Drupal Version: cvs Component: user.module Category: bug reports Priority: minor Assigned to: gordon Reported by: bernardo Updated by: gordon Status: patch (code needs review) Attachment: http://drupal.org/files/issues/user_29.patch (685 bytes) a little tidier patch. gordon Previous comments: ------------------------------------------------------------------------ Thu, 23 Sep 2004 15:14:29 +0000 : bernardo In the user_login function, user_deny is called even if $edit['name'] is empty (for example when the login form is displayed for the first time). I think this performs 2 unnecessary database queries. ------------------------------------------------------------------------ Mon, 08 Aug 2005 21:49:18 +0000 : gordon Attachment: http://drupal.org/files/issues/user_28.patch (691 bytes) here is a small problem that fixes this. It actually reduces the queries by 4 not including and also the actual user authentication.

1 0

[drupal-devel] [bug] Users with 'maintain books' or 'edit own book pages' permission cannot delete book pages
by Boris Mann 09 Aug '05

09 Aug '05

Issue status update for http://drupal.org/node/21559 Post a follow up: http://drupal.org/project/comments/add/21559 Project: Drupal Version: cvs Component: book.module Category: bug reports Priority: normal Assigned to: Anonymous Reported by: clydefrog Updated by: Boris Mann Status: patch (code needs review) puregin: I was talking about this today, so I created a task [1] to remind us to build a generalized node relationship module/schema. [1] http://drupal.org/node/28480 Boris Mann Previous comments: ------------------------------------------------------------------------ Fri, 29 Apr 2005 04:32:19 +0000 : clydefrog Attachment: http://drupal.org/files/issues/delete_own_book_pages.patch (626 bytes) AFAICT, only 'administer nodes' permission allows deletion of book pages. This patch allows users with 'maintain books' or 'edit own book pages' permission to delete book pages. The logic doesn't check if the node has updates pending. Should it? ------------------------------------------------------------------------ Fri, 29 Apr 2005 14:10:22 +0000 : clydefrog It's a patch, duh. ------------------------------------------------------------------------ Fri, 29 Apr 2005 14:21:27 +0000 : rivena If maintain books is a permission that overlaps other permissions, it should say so. For example, /maintain books (create, edit, and delete all book pages)/. While I agree that someone who has been given maintainer status should be able to delete pages, I wonder if 'delete own pages' should be made into a seperate permission. I could see a situation where I would not want someone to be able to delete their own pages. Do these permissions affect things that are not book pages but have been put into books? Anisa. It's /Golden Week/! ------------------------------------------------------------------------ Sat, 30 Apr 2005 03:43:17 +0000 : clydefrog I think it's clear that "maintain books" implies the ability to edit and delete book pages. No other node module provides a "delete own " permission, so it would be inconsistent for book.module to provide it. Can you provide some examples to justify the inconsistency? These permissions do not affect nodes which aren't book pages but are in a book. Those nodes are controlled by their respective type-specific permissions, but only users with "maintain books" permission can add or remove those nodes from books. I believe this is the correct behavior. ------------------------------------------------------------------------ Sat, 30 Apr 2005 11:03:47 +0000 : rivena Hmm... But, you just said that until this patch, it *didn't* include the delete pages permission. Right? So, now you are changing it's behavior, but leaving the wording the same. How will people know? Or does it not matter, because you presume they didn't really know it didn't include delete in the first place? I don't know that I can defend my idea of having a seperate delete own pages permission against a standard of consistency. If all the other edit own permissions include the delete pages permission, then I have no quibbles. :) Except it would be more userfriendly if I knew what the permissions meant before I granted them to someone, without having to look at the code. But I do realize that is completely out of the scope of the patch. :) Where do all these little changes *go*, anyway? The changelog just says things like... refractored the search module. Anisa. ------------------------------------------------------------------------ Sat, 30 Apr 2005 16:04:14 +0000 : clydefrog I believe that "maintain books" permission implies the ability to delete book pages, so this patch fixes that bug. Other node modules include deletion in the "edit own" permission. For example, page.module: <?php if ($op == 'update' || $op == 'delete') { if (user_access('edit own pages') && ($user->uid == $node->uid)) { return TRUE; } } ?> I agree that it would be better if permissions were clearer about what they implied. There was a discussion started by killes [2] a while ago: "One problem with Drupal is the way it handles user permissions. The permissions are attached to the user object but nobody really knows what they do. To find out what a particular permission allows you to do, you often need to have a look at the code. " I don't know if it ever lead to any code. I think the changelog is meant to list the changes since the previous version without too much technical detail. If you want to see every last change, take a look at the cvs messages [3]. [2] http://lists.drupal.org/archives/drupal-devel/2005-04/msg00704.html [3] http://drupal.org/project/cvs/3060 ------------------------------------------------------------------------ Mon, 23 May 2005 17:33:18 +0000 : rivena I've been looking at the book module permissions recently, and did a bit of testing. I wanted to know what was the minimum permissions needed to create a new book. I assumed based on this discussion, and a similar discussion on the drupal support list that this was 'maintain books' and 'administer nodes'. I ran into a problem, then added 'access nodes' to this. So now my user has no other permissions other than 'maintain books', 'administer nodes' and 'access nodes'. You'd think this is more than sufficient to do something simple like create a new book, but in fact, with just these permissions, you do not have access to create a book page. You can edit other pages, and move them into a new book. Once you have create pages permission, you appear to finally have all the permissions needed, edit own pages is not needed. So, since, the point of this patch was that not having delete permission was odd, might I suggest adding create permission to maintain books? ;p Anisa. ------------------------------------------------------------------------ Mon, 08 Aug 2005 10:37:49 +0000 : Bèr Kessels I would say this is a "by design". Maybe we need to rename the two book permissions into: "create books" and "edit own book pages". None of the other nodes can be deleted by the roles with "edit own Foo" permissions, nor should the books. ------------------------------------------------------------------------ Mon, 08 Aug 2005 16:59:30 +0000 : clydefrog Thanks for your review, Ber. story.module [4] allows users with "edit own stories" permission to delete stories. <?php if ($op == 'update' || $op == 'delete') { if (user_access('edit own stories') && ($user->uid == $node->uid)) { return TRUE; } } ?> Therefore it is not inconsistent for the same behavior to apply to book nodes. If "maintain books" really means "create books", then the name should be changed and existing code in book.module needs to be updated. However, I think "maintain books" should allow more than just that. In my situation, I want to be able to give someone the permissions to create books and to create, edit, and delete all book pages. With my patch, "maintain books" and "create book pages" does that. [4] http://cvs.drupal.org/viewcvs/drupal/drupal/modules/story.module?rev=1.167&… ------------------------------------------------------------------------ Mon, 08 Aug 2005 19:16:35 +0000 : Bèr Kessels What i see from story, that is specific to story.module too. We should really settle for a consistent behaviour for *all* nodes, one that is "enforced" from node.module. These things should really not be decided per node-type and per-module. Thats a recipe for a mess. ------------------------------------------------------------------------ Mon, 08 Aug 2005 19:25:27 +0000 : Eric Scouten I agree that the current system is a mess. I disagree that there should be one consistent policy enforced by node.module. Sites may have quite legitimate reasons to have different policies for different node types. ------------------------------------------------------------------------ Mon, 08 Aug 2005 20:35:59 +0000 : Bèr Kessels What i was aiming at is to let node module take care of the following: * edit $node_type * administer $node_type * delete $node_type (but would that not better be in 'administer';: users who can administer can delete) * create $node_type That way the modules need not cre about this. ut they /can/ take care of it, if they wish, and add more rules and permissions. ------------------------------------------------------------------------ Mon, 08 Aug 2005 20:36:32 +0000 : Bèr Kessels oh, and -forgot- : * view $node_type ------------------------------------------------------------------------ Mon, 08 Aug 2005 20:49:31 +0000 : Eric Scouten Thanks for the clarification, Bèr. I was afraid that you meant that there would be a single policy applied to *all* node types, not a policy *per* node type. Now I think we're in complete agreement on this subject. ------------------------------------------------------------------------ Mon, 08 Aug 2005 20:51:13 +0000 : puregin I think it's important to keep a clean separation between content and structure (e.g., what's inside of the node, versus how the node is associated with other nodes) I think that there will likely be other explicit structures in Drupal over time, so these concerns affect more than just book module. Deleting a node from a book is potentially a problem, since this operation affects the structure of a book. We also need to think about other interactions with structure - for example, what happens if someone unpublished a node in a book? Should all children also be unpublished? Should the node contents be hidden, while children remain available in navigation? ------------------------------------------------------------------------ Mon, 08 Aug 2005 21:08:01 +0000 : Bèr Kessels puregrin. I think you are gettnig OT here, Thought it is a very valid (and problematic) issue, It really belongs in a new thread. ------------------------------------------------------------------------ Mon, 08 Aug 2005 22:57:34 +0000 : puregin What I was saying is that whether a user should be able to delete a node from a structure should really be enforced by the module which is responsible for the structure. For example, imagining a simple linear list structure - let's say for a "wish list" - composed of nodes owned by a bunch of different users. In this example, a user could delete his own node with no problem. For the book module, a users node may have children authored by other users. Letting an arbitrary user delete his own node would be, I think, a mistake, since the children would then be orphaned. I suppose one could imagine other modules which rely on a hierarchy, which might have a different policy (e.g., automatically attach children to the parent of the deleted node, if such a node exists). Perhaps eventually two 'tree-based' structure modules might inherit from a tree structure module, each enforcing separate policies regarding node operations. The point is, that one should not have to modify node module when a new structure is defined. Node module should not have to know about how composites of nodes work.

1 0

[drupal-devel] New CVS accounts
by Dries Buytaert 09 Aug '05

09 Aug '05

Hello world, I approved a bunch of CVS accounts. Please welcome our new contributors, and assist them as necessary. -- Dries Buytaert :: http://www.buytaert.net/

2 1

[drupal-devel] [bug] Users with 'maintain books' or 'edit own book pages' permission cannot delete book pages
by puregin 09 Aug '05

09 Aug '05

Issue status update for http://drupal.org/node/21559 Post a follow up: http://drupal.org/project/comments/add/21559 Project: Drupal Version: cvs Component: book.module Category: bug reports Priority: normal Assigned to: Anonymous Reported by: clydefrog Updated by: puregin Status: patch (code needs review) What I was saying is that whether a user should be able to delete a node from a structure should really be enforced by the module which is responsible for the structure. For example, imagining a simple linear list structure - let's say for a "wish list" - composed of nodes owned by a bunch of different users. In this example, a user could delete his own node with no problem. For the book module, a users node may have children authored by other users. Letting an arbitrary user delete his own node would be, I think, a mistake, since the children would then be orphaned. I suppose one could imagine other modules which rely on a hierarchy, which might have a different policy (e.g., automatically attach children to the parent of the deleted node, if such a node exists). Perhaps eventually two 'tree-based' structure modules might inherit from a tree structure module, each enforcing separate policies regarding node operations. The point is, that one should not have to modify node module when a new structure is defined. Node module should not have to know about how composites of nodes work. puregin Previous comments: ------------------------------------------------------------------------ Fri, 29 Apr 2005 04:32:19 +0000 : clydefrog Attachment: http://drupal.org/files/issues/delete_own_book_pages.patch (626 bytes) AFAICT, only 'administer nodes' permission allows deletion of book pages. This patch allows users with 'maintain books' or 'edit own book pages' permission to delete book pages. The logic doesn't check if the node has updates pending. Should it? ------------------------------------------------------------------------ Fri, 29 Apr 2005 14:10:22 +0000 : clydefrog It's a patch, duh. ------------------------------------------------------------------------ Fri, 29 Apr 2005 14:21:27 +0000 : rivena If maintain books is a permission that overlaps other permissions, it should say so. For example, /maintain books (create, edit, and delete all book pages)/. While I agree that someone who has been given maintainer status should be able to delete pages, I wonder if 'delete own pages' should be made into a seperate permission. I could see a situation where I would not want someone to be able to delete their own pages. Do these permissions affect things that are not book pages but have been put into books? Anisa. It's /Golden Week/! ------------------------------------------------------------------------ Sat, 30 Apr 2005 03:43:17 +0000 : clydefrog I think it's clear that "maintain books" implies the ability to edit and delete book pages. No other node module provides a "delete own " permission, so it would be inconsistent for book.module to provide it. Can you provide some examples to justify the inconsistency? These permissions do not affect nodes which aren't book pages but are in a book. Those nodes are controlled by their respective type-specific permissions, but only users with "maintain books" permission can add or remove those nodes from books. I believe this is the correct behavior. ------------------------------------------------------------------------ Sat, 30 Apr 2005 11:03:47 +0000 : rivena Hmm... But, you just said that until this patch, it *didn't* include the delete pages permission. Right? So, now you are changing it's behavior, but leaving the wording the same. How will people know? Or does it not matter, because you presume they didn't really know it didn't include delete in the first place? I don't know that I can defend my idea of having a seperate delete own pages permission against a standard of consistency. If all the other edit own permissions include the delete pages permission, then I have no quibbles. :) Except it would be more userfriendly if I knew what the permissions meant before I granted them to someone, without having to look at the code. But I do realize that is completely out of the scope of the patch. :) Where do all these little changes *go*, anyway? The changelog just says things like... refractored the search module. Anisa. ------------------------------------------------------------------------ Sat, 30 Apr 2005 16:04:14 +0000 : clydefrog I believe that "maintain books" permission implies the ability to delete book pages, so this patch fixes that bug. Other node modules include deletion in the "edit own" permission. For example, page.module: <?php if ($op == 'update' || $op == 'delete') { if (user_access('edit own pages') && ($user->uid == $node->uid)) { return TRUE; } } ?> I agree that it would be better if permissions were clearer about what they implied. There was a discussion started by killes [1] a while ago: "One problem with Drupal is the way it handles user permissions. The permissions are attached to the user object but nobody really knows what they do. To find out what a particular permission allows you to do, you often need to have a look at the code. " I don't know if it ever lead to any code. I think the changelog is meant to list the changes since the previous version without too much technical detail. If you want to see every last change, take a look at the cvs messages [2]. [1] http://lists.drupal.org/archives/drupal-devel/2005-04/msg00704.html [2] http://drupal.org/project/cvs/3060 ------------------------------------------------------------------------ Mon, 23 May 2005 17:33:18 +0000 : rivena I've been looking at the book module permissions recently, and did a bit of testing. I wanted to know what was the minimum permissions needed to create a new book. I assumed based on this discussion, and a similar discussion on the drupal support list that this was 'maintain books' and 'administer nodes'. I ran into a problem, then added 'access nodes' to this. So now my user has no other permissions other than 'maintain books', 'administer nodes' and 'access nodes'. You'd think this is more than sufficient to do something simple like create a new book, but in fact, with just these permissions, you do not have access to create a book page. You can edit other pages, and move them into a new book. Once you have create pages permission, you appear to finally have all the permissions needed, edit own pages is not needed. So, since, the point of this patch was that not having delete permission was odd, might I suggest adding create permission to maintain books? ;p Anisa. ------------------------------------------------------------------------ Mon, 08 Aug 2005 10:37:49 +0000 : Bèr Kessels I would say this is a "by design". Maybe we need to rename the two book permissions into: "create books" and "edit own book pages". None of the other nodes can be deleted by the roles with "edit own Foo" permissions, nor should the books. ------------------------------------------------------------------------ Mon, 08 Aug 2005 16:59:30 +0000 : clydefrog Thanks for your review, Ber. story.module [3] allows users with "edit own stories" permission to delete stories. <?php if ($op == 'update' || $op == 'delete') { if (user_access('edit own stories') && ($user->uid == $node->uid)) { return TRUE; } } ?> Therefore it is not inconsistent for the same behavior to apply to book nodes. If "maintain books" really means "create books", then the name should be changed and existing code in book.module needs to be updated. However, I think "maintain books" should allow more than just that. In my situation, I want to be able to give someone the permissions to create books and to create, edit, and delete all book pages. With my patch, "maintain books" and "create book pages" does that. [3] http://cvs.drupal.org/viewcvs/drupal/drupal/modules/story.module?rev=1.167&… ------------------------------------------------------------------------ Mon, 08 Aug 2005 19:16:35 +0000 : Bèr Kessels What i see from story, that is specific to story.module too. We should really settle for a consistent behaviour for *all* nodes, one that is "enforced" from node.module. These things should really not be decided per node-type and per-module. Thats a recipe for a mess. ------------------------------------------------------------------------ Mon, 08 Aug 2005 19:25:27 +0000 : Eric Scouten I agree that the current system is a mess. I disagree that there should be one consistent policy enforced by node.module. Sites may have quite legitimate reasons to have different policies for different node types. ------------------------------------------------------------------------ Mon, 08 Aug 2005 20:35:59 +0000 : Bèr Kessels What i was aiming at is to let node module take care of the following: * edit $node_type * administer $node_type * delete $node_type (but would that not better be in 'administer';: users who can administer can delete) * create $node_type That way the modules need not cre about this. ut they /can/ take care of it, if they wish, and add more rules and permissions. ------------------------------------------------------------------------ Mon, 08 Aug 2005 20:36:32 +0000 : Bèr Kessels oh, and -forgot- : * view $node_type ------------------------------------------------------------------------ Mon, 08 Aug 2005 20:49:31 +0000 : Eric Scouten Thanks for the clarification, Bèr. I was afraid that you meant that there would be a single policy applied to *all* node types, not a policy *per* node type. Now I think we're in complete agreement on this subject. ------------------------------------------------------------------------ Mon, 08 Aug 2005 20:51:13 +0000 : puregin I think it's important to keep a clean separation between content and structure (e.g., what's inside of the node, versus how the node is associated with other nodes) I think that there will likely be other explicit structures in Drupal over time, so these concerns affect more than just book module. Deleting a node from a book is potentially a problem, since this operation affects the structure of a book. We also need to think about other interactions with structure - for example, what happens if someone unpublished a node in a book? Should all children also be unpublished? Should the node contents be hidden, while children remain available in navigation? ------------------------------------------------------------------------ Mon, 08 Aug 2005 21:08:01 +0000 : Bèr Kessels puregrin. I think you are gettnig OT here, Thought it is a very valid (and problematic) issue, It really belongs in a new thread.

1 0

[drupal-devel] [bug] user_login calls user_deny with empty variable
by gordon 08 Aug '05

08 Aug '05

Issue status update for http://drupal.org/node/11031 Post a follow up: http://drupal.org/project/comments/add/11031 Project: Drupal Version: cvs Component: user.module Category: bug reports Priority: minor -Assigned to: Anonymous +Assigned to: gordon Reported by: bernardo Updated by: gordon -Status: active +Status: patch (code needs review) Attachment: http://drupal.org/files/issues/user_28.patch (691 bytes) here is a small problem that fixes this. It actually reduces the queries by 4 not including and also the actual user authentication. gordon Previous comments: ------------------------------------------------------------------------ Thu, 23 Sep 2004 15:14:29 +0000 : bernardo In the user_login function, user_deny is called even if $edit['name'] is empty (for example when the login form is displayed for the first time). I think this performs 2 unnecessary database queries.

1 0

[drupal-devel] [bug] Some URLs get ?PHPSESSID added to them
by kbahey 08 Aug '05

08 Aug '05

Issue status update for http://drupal.org/node/4109 Post a follow up: http://drupal.org/project/comments/add/4109 Project: Drupal Version: cvs Component: base system Category: bug reports Priority: normal Assigned to: Eric Scouten Reported by: Eric Scouten Updated by: kbahey -Status: patch (ready to be committed) +Status: patch (code needs review) Attachment: http://drupal.org/files/issues/common.inc-trans_sid.patch (759 bytes) Actually, since the code in common.inc should not fiddle with PHPSESSID in URL in the first place, why not junk that whole section of the code altogether? A patch for common.inc is attached that removes this whole section. As Moshe said, this is cruft. kbahey Previous comments: ------------------------------------------------------------------------ Thu, 13 Nov 2003 00:22:39 +0000 : Eric Scouten Some URLs get ?PHPSESSID=(whatever) added to them. I find this distracting and unnecessary for anonymous users. Submitting patch #147 for consideration to remove this issue. ------------------------------------------------------------------------ Fri, 05 Dec 2003 14:39:25 +0000 : Kjartan This is a local PHP configuration issue. Don't see it as something Drupal should mess with, the less PHP settings Drupal messes with the better. ------------------------------------------------------------------------ Fri, 05 Dec 2003 15:11:33 +0000 : killes(a)www.drop.org Could you elaborate which config setting one has to tweak? ------------------------------------------------------------------------ Fri, 05 Dec 2003 15:18:11 +0000 : Anonymous php_flag session.use_trans_sid off php_flag session.use_only_cookies on The first will disable the automatic addition of the session ID to internal URLs, the second will remove support for passing session IDs in URLs. See http://php.net/session [1] for more info. BTW the above settings are for Apache, in PHP.ini you should not add the php_flag part, and you should use = inbetween the name and value... [1] http://php.net/session ------------------------------------------------------------------------ Mon, 07 Jun 2004 22:42:06 +0000 : shane It seems that Drupal (via the .htaccess file) already "messes with" a bunch of PHP settings. For example, my Drupal install "came with" these messed with settings: php_value register_globals 0 php_value track_vars 1 php_value short_open_tag 1 php_value magic_quotes_gpc 0 php_value magic_quotes_runtime 0 php_value magic_quotes_sybase 0 php_value arg_separator.output "&" php_value session.cache_expire 200000 php_value session.gc_maxlifetime 200000 php_value session.cookie_lifetime 2000000 php_value session.auto_start 0 php_value session.save_handler user php_value session.cache_limiter none php_value allow_call_time_pass_reference On I'm not sure that adding two more "php_value" flags to the stock .htaccess file is going to arguably mess with more than already is. Personally I find it extremely un-professional and very un-clean to have funky PHPSESSID url strings tacked onto a majority of my URLs. I've worked extremely hard to implement clean URLs (above and beyond the already great "clean url" feature of Drupal - I don't like them being turned into muck. I'd vote that these two flags be added to the .htaccess file that is distributed with Drupal. However, I'm uncertain of the overall true impact of making these changes, so barring an technical reason - I would absolutely request this as a feature. I've implemented these in many of my production sites already (just today) - we'll see what impact they have. ------------------------------------------------------------------------ Tue, 08 Jun 2004 02:56:00 +0000 : marky It's a performance thing. You need to remember that while not everyone has edit access to php.ini, those of us that do control our own servers see no need for .htaccess directives that set php options. >From the Apache docs [2]: "Apache will look in every directory for .htaccess files. Thus, permitting .htaccess files causes a performance hit, whether or not you actually even use them! Also, the .htaccess file is loaded every time a document is requested." Ok, I'm lucky in that I control my own server, so I can effectively remove the .htaccess file and move its config options to php.ini and the vhost section of httpd.conf, thus speeding up the responsiveness of my server. But if you don't have that option there's nothing wrong with adding those directives to your directory .htaccess file. That's what it's for - it allows directory specific config options for those that don't have access to their server config. As Kjartan said, "the less PHP settings Drupal messes with the better". I've got to agree - as far as I'm concerned it's already big enough. The "php_value short_open_tag 1", "php_value session.auto_start 0" and "php_value session.cache_limiter none" for example serve little purpose, as they mirror the installation default settings (apart from the last, where the available options are: nocache, private or public). If you don't have the choice, use .htaccess, and add whatever directives you see fit. But please remember that not everyone uses it. Anyway, just my 2 beads. [2] http://httpd.apache.org/docs-2.0/howto/htaccess.html#when ------------------------------------------------------------------------ Tue, 08 Jun 2004 04:38:32 +0000 : cel4145 Why not add them, but commented out with an explanation? When I encountered this problem, I had to spend a little time on drupal.org finding the solution. However, I am familiar with the .htaccess file from doing installs. If it had been included there commented out, I would have probably remembered it. After all, this is much the same way that httpd.conf functions with it's many, optional, commented out configuration options. Just makes it easier when a non-standard configuration option has to be used. Besides, I've also seen many Drupal sites with this problem. I suspect that some of those sites would have implemented these flags if they had been an option in the .htaccess file. And it looks bad. People that visit the sites that don't know any better will assume it's Drupal, not the lack of proper PHP settings. ------------------------------------------------------------------------ Tue, 08 Jun 2004 10:39:27 +0000 : killes(a)www.drop.org It wouldn't hurt if Drupal had a better documented and possibly extended .htaccess file. Those who whish not to use it should really not care... session.save_handler = user should probably included as well. ------------------------------------------------------------------------ Tue, 08 Jun 2004 12:17:06 +0000 : robertDouglass If you use these settings: php_flag session.use_trans_sid off php_flag session.use_only_cookies on won't it make it so that people with cookies disabled cannot have session state? And if the PHPSESSID is in the URL, isn't that an indication that the browser being used refused to take a cookie? I'd really like clarification on this, if anyone knows the answers. ------------------------------------------------------------------------ Tue, 27 Jul 2004 13:25:46 +0000 : JonBob The original issue is definitely by design, as these URLs need the session ID to preserve the session when running under the given PHP settings. A better-documented .htaccess is a different issue entirely. ------------------------------------------------------------------------ Thu, 19 Aug 2004 18:04:13 +0000 : gtoddv It hasn't been mentioned in this thread or anywhere else that I can find, but this session ID issue breaks validation. I don't know is this will cause accessibility issues too. I have tried the .htaccess fix and that doesn't change anything. ------------------------------------------------------------------------ Sat, 06 Nov 2004 07:17:21 +0000 : wazdog The non-validating problem is another problem with your server setup. PHP can be setup to used encoded &'s or not. Your server isn't if the url isn't validating. You'd have to talk to your server admin to fix that. ------------------------------------------------------------------------ Thu, 30 Dec 2004 20:32:35 +0000 : m_freeman2004 This is what you have to include in the .htaccess file: # Fix for ?PHPSESSID in clean URLs php_value session.use_trans_sid 0 php_value session.use_only_cookies 1 # End of fix ------------------------------------------------------------------------ Wed, 01 Jun 2005 19:07:53 +0000 : dr05 The code in .htaccess file: # Fix for ?PHPSESSID in clean URLs php_value session.use_trans_sid 0 php_value session.use_only_cookies 1 # End of fix don't work! Error: "Internal Server Error!" ------------------------------------------------------------------------ Thu, 14 Jul 2005 15:25:22 +0000 : jasonwhat Anyone else looking into this? Can one get rid of the id added onto the url but still keep the sessions going? ------------------------------------------------------------------------ Mon, 25 Jul 2005 06:16:22 +0000 : Eric Scouten dr05, what version of Apache and what OS are you using? I've been using this same patch with no problems on Mac OS X, Linux, and FreeBSD-based servers using both Apache 1.3.x and 2.0.x. Attached patch will apply these changes to .htaccess. (IMHO, this is useful for the majority of users; those who are smart enough to understand the performance issues can comment out these lines.) ------------------------------------------------------------------------ Mon, 25 Jul 2005 06:57:18 +0000 : clydefrog You forgot the patch, Eric. ------------------------------------------------------------------------ Mon, 25 Jul 2005 08:02:26 +0000 : Eric Scouten Attachment: http://drupal.org/files/issues/htaccess-no-session-ids.patch (383 bytes) D'oh! ------------------------------------------------------------------------ Mon, 25 Jul 2005 08:45:48 +0000 : kbahey There are cases where no parameter will fix this issue. On shared hosting systems where you do not have access to php.ini, and the hosting company has setup things in a way that they cannot be overriden. I found this out the hard way, after lobbying for the settings to be included in the init_set() statements in the settings.php prior to 4.6 being released. They are simply ignored on some hosts. On other hosts, you need to have a php.ini of your own, if suphp or SuExec is being used. In some cases, the only way to make things work is to compile your own PHP as a CGI (detailed writeup) [3]. This can have disadvantages too, such some performance impact, as well as excessive CPU utilization (hosting company may not like it). [3] http://baheyeldin.com/drupal/how-to-get-rid-of-phpsessid-in-drupal-and-othe… ------------------------------------------------------------------------ Mon, 25 Jul 2005 08:49:31 +0000 : kbahey Here is some references: issue 21170 [4], and a forum discussion [5]. [4] http://drupal.org/node/21170 [5] http://drupal.org/node/17947 ------------------------------------------------------------------------ Sat, 30 Jul 2005 16:24:47 +0000 : ufku in my case (shared hosting, ini_set() disabled) a php.ini file with only this line in it solved the problem. url_rewriter.tags = '' you should place the php.ini in your site's base directory ------------------------------------------------------------------------ Sun, 07 Aug 2005 12:18:31 +0000 : djnz Attachment: http://drupal.org/files/issues/rewrite_tags.patch (468 bytes) Drupal already attempts to change a number of values using ini_set() in settings.php, including session.use_trans_sid which should cure this problem. Unfortunately as already noted elsewhere [6], session.use_trans_sid only worked in ini_set() up to PHP 4.2.3. Rather than change the whole philosophy and change this setting in .htaccess or php.ini (neither of which work for many hosted set-ups), surely the answer is to correct settings.php as described The attached patch does just that - it is only one line. Note that session ids are still appended to the url when redirecting by Drupal eg after a POST. This is also easily patched [7], but as it is done deliberately (why?) I have left it alone. [6] http://drupal.org/node/17947#comment-36339 [7] http://drupal.org/node/25852#comment-45111 ------------------------------------------------------------------------ Sun, 07 Aug 2005 13:26:17 +0000 : kbahey PHPSESSIDs in the URL are a) ugly, b) negatively affects search engine indexing, and most importantly, c) is a security risk that can enable a malicious user to hijack your session. We should work on a solution to eliminate this once and for all, and for all configurations. The current approach (putting some PHP parameters in settings.php, or instructing users to change them in .htaccess or php.ini) has severe shorcomings: Whatever we do, there are bound to be some configurations that no amount of PHP parameters will cure. Some hosts deny their users the ability to change some or all of the above parameters. I discovered this the painful way with many hosts I have used, and friends and clients I have helped. I think we need to rethink this whole session thing, in a way that we are not dependant on idiosyncracies of various hosts and PHP configurations. Perhaps attempt to set a DRUPAL cookie and not maintain a session unless we are able to set it. This way we know that we will not have a PHPSESSID in the URL. If we are not able to set the cookie, then we allow only anonymous access that is sessionless.. I am not sure if that will work or not: what other options do we have here? ------------------------------------------------------------------------ Sun, 07 Aug 2005 22:02:19 +0000 : djnz I am convinced that ini_set('url_rewriter.tags','') is the answer. I find it hard to believe that any host would go to the trouble to disable this (not just a matter of setting compile time paramaters, you would have to hack the PHP source; and why would they?) - certainly no cPanel host I have used does. The reason the code currently in settings.php doesn't work is not because hosts disable it, it is because PHP does not support it. RTFM. ------------------------------------------------------------------------ Sun, 07 Aug 2005 22:23:59 +0000 : djnz Attachment: http://drupal.org/files/issues/trans_sid.patch (1.4 KB) After further thought, there really is no point in appending the SID to a redirect url, so the attached patch solves this problem too. If anyone has a problem with session IDs in URLs after applying this patch, I'll offer them free hosting myself. Maybe. ;) ------------------------------------------------------------------------ Mon, 08 Aug 2005 03:14:41 +0000 : kbahey You are right about the fact that the use of PHPSESSID in the URL on a redirect is totally unnecessary. That line should be removed from the base code altogether, there is no excuse for it. As for url_rewriter.tags, I am not 100% sure, but I tried it with someone who was facing difficulties, and it did not work as far as I can recall. In any case, your solution should go in base. It cannot hurt, unless some host has disabled it. ------------------------------------------------------------------------ Mon, 08 Aug 2005 20:09:40 +0000 : moshe weitzman yeah, i think that is cruft.

1 0

[drupal-devel] [bug] Users with 'maintain books' or 'edit own book pages' permission cannot delete book pages
by Bèr Kessels 08 Aug '05

08 Aug '05

Issue status update for http://drupal.org/node/21559 Post a follow up: http://drupal.org/project/comments/add/21559 Project: Drupal Version: cvs Component: book.module Category: bug reports Priority: normal Assigned to: Anonymous Reported by: clydefrog Updated by: Bèr Kessels Status: patch (code needs review) puregrin. I think you are gettnig OT here, Thought it is a very valid (and problematic) issue, It really belongs in a new thread. Bèr Kessels Previous comments: ------------------------------------------------------------------------ Fri, 29 Apr 2005 04:32:19 +0000 : clydefrog Attachment: http://drupal.org/files/issues/delete_own_book_pages.patch (626 bytes) AFAICT, only 'administer nodes' permission allows deletion of book pages. This patch allows users with 'maintain books' or 'edit own book pages' permission to delete book pages. The logic doesn't check if the node has updates pending. Should it? ------------------------------------------------------------------------ Fri, 29 Apr 2005 14:10:22 +0000 : clydefrog It's a patch, duh. ------------------------------------------------------------------------ Fri, 29 Apr 2005 14:21:27 +0000 : rivena If maintain books is a permission that overlaps other permissions, it should say so. For example, /maintain books (create, edit, and delete all book pages)/. While I agree that someone who has been given maintainer status should be able to delete pages, I wonder if 'delete own pages' should be made into a seperate permission. I could see a situation where I would not want someone to be able to delete their own pages. Do these permissions affect things that are not book pages but have been put into books? Anisa. It's /Golden Week/! ------------------------------------------------------------------------ Sat, 30 Apr 2005 03:43:17 +0000 : clydefrog I think it's clear that "maintain books" implies the ability to edit and delete book pages. No other node module provides a "delete own " permission, so it would be inconsistent for book.module to provide it. Can you provide some examples to justify the inconsistency? These permissions do not affect nodes which aren't book pages but are in a book. Those nodes are controlled by their respective type-specific permissions, but only users with "maintain books" permission can add or remove those nodes from books. I believe this is the correct behavior. ------------------------------------------------------------------------ Sat, 30 Apr 2005 11:03:47 +0000 : rivena Hmm... But, you just said that until this patch, it *didn't* include the delete pages permission. Right? So, now you are changing it's behavior, but leaving the wording the same. How will people know? Or does it not matter, because you presume they didn't really know it didn't include delete in the first place? I don't know that I can defend my idea of having a seperate delete own pages permission against a standard of consistency. If all the other edit own permissions include the delete pages permission, then I have no quibbles. :) Except it would be more userfriendly if I knew what the permissions meant before I granted them to someone, without having to look at the code. But I do realize that is completely out of the scope of the patch. :) Where do all these little changes *go*, anyway? The changelog just says things like... refractored the search module. Anisa. ------------------------------------------------------------------------ Sat, 30 Apr 2005 16:04:14 +0000 : clydefrog I believe that "maintain books" permission implies the ability to delete book pages, so this patch fixes that bug. Other node modules include deletion in the "edit own" permission. For example, page.module: <?php if ($op == 'update' || $op == 'delete') { if (user_access('edit own pages') && ($user->uid == $node->uid)) { return TRUE; } } ?> I agree that it would be better if permissions were clearer about what they implied. There was a discussion started by killes [1] a while ago: "One problem with Drupal is the way it handles user permissions. The permissions are attached to the user object but nobody really knows what they do. To find out what a particular permission allows you to do, you often need to have a look at the code. " I don't know if it ever lead to any code. I think the changelog is meant to list the changes since the previous version without too much technical detail. If you want to see every last change, take a look at the cvs messages [2]. [1] http://lists.drupal.org/archives/drupal-devel/2005-04/msg00704.html [2] http://drupal.org/project/cvs/3060 ------------------------------------------------------------------------ Mon, 23 May 2005 17:33:18 +0000 : rivena I've been looking at the book module permissions recently, and did a bit of testing. I wanted to know what was the minimum permissions needed to create a new book. I assumed based on this discussion, and a similar discussion on the drupal support list that this was 'maintain books' and 'administer nodes'. I ran into a problem, then added 'access nodes' to this. So now my user has no other permissions other than 'maintain books', 'administer nodes' and 'access nodes'. You'd think this is more than sufficient to do something simple like create a new book, but in fact, with just these permissions, you do not have access to create a book page. You can edit other pages, and move them into a new book. Once you have create pages permission, you appear to finally have all the permissions needed, edit own pages is not needed. So, since, the point of this patch was that not having delete permission was odd, might I suggest adding create permission to maintain books? ;p Anisa. ------------------------------------------------------------------------ Mon, 08 Aug 2005 10:37:49 +0000 : Bèr Kessels I would say this is a "by design". Maybe we need to rename the two book permissions into: "create books" and "edit own book pages". None of the other nodes can be deleted by the roles with "edit own Foo" permissions, nor should the books. ------------------------------------------------------------------------ Mon, 08 Aug 2005 16:59:30 +0000 : clydefrog Thanks for your review, Ber. story.module [3] allows users with "edit own stories" permission to delete stories. <?php if ($op == 'update' || $op == 'delete') { if (user_access('edit own stories') && ($user->uid == $node->uid)) { return TRUE; } } ?> Therefore it is not inconsistent for the same behavior to apply to book nodes. If "maintain books" really means "create books", then the name should be changed and existing code in book.module needs to be updated. However, I think "maintain books" should allow more than just that. In my situation, I want to be able to give someone the permissions to create books and to create, edit, and delete all book pages. With my patch, "maintain books" and "create book pages" does that. [3] http://cvs.drupal.org/viewcvs/drupal/drupal/modules/story.module?rev=1.167&… ------------------------------------------------------------------------ Mon, 08 Aug 2005 19:16:35 +0000 : Bèr Kessels What i see from story, that is specific to story.module too. We should really settle for a consistent behaviour for *all* nodes, one that is "enforced" from node.module. These things should really not be decided per node-type and per-module. Thats a recipe for a mess. ------------------------------------------------------------------------ Mon, 08 Aug 2005 19:25:27 +0000 : Eric Scouten I agree that the current system is a mess. I disagree that there should be one consistent policy enforced by node.module. Sites may have quite legitimate reasons to have different policies for different node types. ------------------------------------------------------------------------ Mon, 08 Aug 2005 20:35:59 +0000 : Bèr Kessels What i was aiming at is to let node module take care of the following: * edit $node_type * administer $node_type * delete $node_type (but would that not better be in 'administer';: users who can administer can delete) * create $node_type That way the modules need not cre about this. ut they /can/ take care of it, if they wish, and add more rules and permissions. ------------------------------------------------------------------------ Mon, 08 Aug 2005 20:36:32 +0000 : Bèr Kessels oh, and -forgot- : * view $node_type ------------------------------------------------------------------------ Mon, 08 Aug 2005 20:49:31 +0000 : Eric Scouten Thanks for the clarification, Bèr. I was afraid that you meant that there would be a single policy applied to *all* node types, not a policy *per* node type. Now I think we're in complete agreement on this subject. ------------------------------------------------------------------------ Mon, 08 Aug 2005 20:51:13 +0000 : puregin I think it's important to keep a clean separation between content and structure (e.g., what's inside of the node, versus how the node is associated with other nodes) I think that there will likely be other explicit structures in Drupal over time, so these concerns affect more than just book module. Deleting a node from a book is potentially a problem, since this operation affects the structure of a book. We also need to think about other interactions with structure - for example, what happens if someone unpublished a node in a book? Should all children also be unpublished? Should the node contents be hidden, while children remain available in navigation?

1 0

[drupal-devel] [bug] Users with 'maintain books' or 'edit own book pages' permission cannot delete book pages
by puregin 08 Aug '05

08 Aug '05

Issue status update for http://drupal.org/node/21559 Post a follow up: http://drupal.org/project/comments/add/21559 Project: Drupal Version: cvs Component: book.module Category: bug reports Priority: normal Assigned to: Anonymous Reported by: clydefrog Updated by: puregin Status: patch (code needs review) I think it's important to keep a clean separation between content and structure (e.g., what's inside of the node, versus how the node is associated with other nodes) I think that there will likely be other explicit structures in Drupal over time, so these concerns affect more than just book module. Deleting a node from a book is potentially a problem, since this operation affects the structure of a book. We also need to think about other interactions with structure - for example, what happens if someone unpublished a node in a book? Should all children also be unpublished? Should the node contents be hidden, while children remain available in navigation? puregin Previous comments: ------------------------------------------------------------------------ Fri, 29 Apr 2005 04:32:19 +0000 : clydefrog Attachment: http://drupal.org/files/issues/delete_own_book_pages.patch (626 bytes) AFAICT, only 'administer nodes' permission allows deletion of book pages. This patch allows users with 'maintain books' or 'edit own book pages' permission to delete book pages. The logic doesn't check if the node has updates pending. Should it? ------------------------------------------------------------------------ Fri, 29 Apr 2005 14:10:22 +0000 : clydefrog It's a patch, duh. ------------------------------------------------------------------------ Fri, 29 Apr 2005 14:21:27 +0000 : rivena If maintain books is a permission that overlaps other permissions, it should say so. For example, /maintain books (create, edit, and delete all book pages)/. While I agree that someone who has been given maintainer status should be able to delete pages, I wonder if 'delete own pages' should be made into a seperate permission. I could see a situation where I would not want someone to be able to delete their own pages. Do these permissions affect things that are not book pages but have been put into books? Anisa. It's /Golden Week/! ------------------------------------------------------------------------ Sat, 30 Apr 2005 03:43:17 +0000 : clydefrog I think it's clear that "maintain books" implies the ability to edit and delete book pages. No other node module provides a "delete own " permission, so it would be inconsistent for book.module to provide it. Can you provide some examples to justify the inconsistency? These permissions do not affect nodes which aren't book pages but are in a book. Those nodes are controlled by their respective type-specific permissions, but only users with "maintain books" permission can add or remove those nodes from books. I believe this is the correct behavior. ------------------------------------------------------------------------ Sat, 30 Apr 2005 11:03:47 +0000 : rivena Hmm... But, you just said that until this patch, it *didn't* include the delete pages permission. Right? So, now you are changing it's behavior, but leaving the wording the same. How will people know? Or does it not matter, because you presume they didn't really know it didn't include delete in the first place? I don't know that I can defend my idea of having a seperate delete own pages permission against a standard of consistency. If all the other edit own permissions include the delete pages permission, then I have no quibbles. :) Except it would be more userfriendly if I knew what the permissions meant before I granted them to someone, without having to look at the code. But I do realize that is completely out of the scope of the patch. :) Where do all these little changes *go*, anyway? The changelog just says things like... refractored the search module. Anisa. ------------------------------------------------------------------------ Sat, 30 Apr 2005 16:04:14 +0000 : clydefrog I believe that "maintain books" permission implies the ability to delete book pages, so this patch fixes that bug. Other node modules include deletion in the "edit own" permission. For example, page.module: <?php if ($op == 'update' || $op == 'delete') { if (user_access('edit own pages') && ($user->uid == $node->uid)) { return TRUE; } } ?> I agree that it would be better if permissions were clearer about what they implied. There was a discussion started by killes [1] a while ago: "One problem with Drupal is the way it handles user permissions. The permissions are attached to the user object but nobody really knows what they do. To find out what a particular permission allows you to do, you often need to have a look at the code. " I don't know if it ever lead to any code. I think the changelog is meant to list the changes since the previous version without too much technical detail. If you want to see every last change, take a look at the cvs messages [2]. [1] http://lists.drupal.org/archives/drupal-devel/2005-04/msg00704.html [2] http://drupal.org/project/cvs/3060 ------------------------------------------------------------------------ Mon, 23 May 2005 17:33:18 +0000 : rivena I've been looking at the book module permissions recently, and did a bit of testing. I wanted to know what was the minimum permissions needed to create a new book. I assumed based on this discussion, and a similar discussion on the drupal support list that this was 'maintain books' and 'administer nodes'. I ran into a problem, then added 'access nodes' to this. So now my user has no other permissions other than 'maintain books', 'administer nodes' and 'access nodes'. You'd think this is more than sufficient to do something simple like create a new book, but in fact, with just these permissions, you do not have access to create a book page. You can edit other pages, and move them into a new book. Once you have create pages permission, you appear to finally have all the permissions needed, edit own pages is not needed. So, since, the point of this patch was that not having delete permission was odd, might I suggest adding create permission to maintain books? ;p Anisa. ------------------------------------------------------------------------ Mon, 08 Aug 2005 10:37:49 +0000 : Bèr Kessels I would say this is a "by design". Maybe we need to rename the two book permissions into: "create books" and "edit own book pages". None of the other nodes can be deleted by the roles with "edit own Foo" permissions, nor should the books. ------------------------------------------------------------------------ Mon, 08 Aug 2005 16:59:30 +0000 : clydefrog Thanks for your review, Ber. story.module [3] allows users with "edit own stories" permission to delete stories. <?php if ($op == 'update' || $op == 'delete') { if (user_access('edit own stories') && ($user->uid == $node->uid)) { return TRUE; } } ?> Therefore it is not inconsistent for the same behavior to apply to book nodes. If "maintain books" really means "create books", then the name should be changed and existing code in book.module needs to be updated. However, I think "maintain books" should allow more than just that. In my situation, I want to be able to give someone the permissions to create books and to create, edit, and delete all book pages. With my patch, "maintain books" and "create book pages" does that. [3] http://cvs.drupal.org/viewcvs/drupal/drupal/modules/story.module?rev=1.167&… ------------------------------------------------------------------------ Mon, 08 Aug 2005 19:16:35 +0000 : Bèr Kessels What i see from story, that is specific to story.module too. We should really settle for a consistent behaviour for *all* nodes, one that is "enforced" from node.module. These things should really not be decided per node-type and per-module. Thats a recipe for a mess. ------------------------------------------------------------------------ Mon, 08 Aug 2005 19:25:27 +0000 : Eric Scouten I agree that the current system is a mess. I disagree that there should be one consistent policy enforced by node.module. Sites may have quite legitimate reasons to have different policies for different node types. ------------------------------------------------------------------------ Mon, 08 Aug 2005 20:35:59 +0000 : Bèr Kessels What i was aiming at is to let node module take care of the following: * edit $node_type * administer $node_type * delete $node_type (but would that not better be in 'administer';: users who can administer can delete) * create $node_type That way the modules need not cre about this. ut they /can/ take care of it, if they wish, and add more rules and permissions. ------------------------------------------------------------------------ Mon, 08 Aug 2005 20:36:32 +0000 : Bèr Kessels oh, and -forgot- : * view $node_type ------------------------------------------------------------------------ Mon, 08 Aug 2005 20:49:31 +0000 : Eric Scouten Thanks for the clarification, Bèr. I was afraid that you meant that there would be a single policy applied to *all* node types, not a policy *per* node type. Now I think we're in complete agreement on this subject.

1 0

[drupal-devel] [bug] Users with 'maintain books' or 'edit own book pages' permission cannot delete book pages
by Eric Scouten 08 Aug '05

08 Aug '05

Issue status update for http://drupal.org/node/21559 Post a follow up: http://drupal.org/project/comments/add/21559 Project: Drupal Version: cvs Component: book.module Category: bug reports Priority: normal Assigned to: Anonymous Reported by: clydefrog Updated by: Eric Scouten Status: patch (code needs review) Thanks for the clarification, Bèr. I was afraid that you meant that there would be a single policy applied to *all* node types, not a policy *per* node type. Now I think we're in complete agreement on this subject. Eric Scouten Previous comments: ------------------------------------------------------------------------ Fri, 29 Apr 2005 04:32:19 +0000 : clydefrog Attachment: http://drupal.org/files/issues/delete_own_book_pages.patch (626 bytes) AFAICT, only 'administer nodes' permission allows deletion of book pages. This patch allows users with 'maintain books' or 'edit own book pages' permission to delete book pages. The logic doesn't check if the node has updates pending. Should it? ------------------------------------------------------------------------ Fri, 29 Apr 2005 14:10:22 +0000 : clydefrog It's a patch, duh. ------------------------------------------------------------------------ Fri, 29 Apr 2005 14:21:27 +0000 : rivena If maintain books is a permission that overlaps other permissions, it should say so. For example, /maintain books (create, edit, and delete all book pages)/. While I agree that someone who has been given maintainer status should be able to delete pages, I wonder if 'delete own pages' should be made into a seperate permission. I could see a situation where I would not want someone to be able to delete their own pages. Do these permissions affect things that are not book pages but have been put into books? Anisa. It's /Golden Week/! ------------------------------------------------------------------------ Sat, 30 Apr 2005 03:43:17 +0000 : clydefrog I think it's clear that "maintain books" implies the ability to edit and delete book pages. No other node module provides a "delete own " permission, so it would be inconsistent for book.module to provide it. Can you provide some examples to justify the inconsistency? These permissions do not affect nodes which aren't book pages but are in a book. Those nodes are controlled by their respective type-specific permissions, but only users with "maintain books" permission can add or remove those nodes from books. I believe this is the correct behavior. ------------------------------------------------------------------------ Sat, 30 Apr 2005 11:03:47 +0000 : rivena Hmm... But, you just said that until this patch, it *didn't* include the delete pages permission. Right? So, now you are changing it's behavior, but leaving the wording the same. How will people know? Or does it not matter, because you presume they didn't really know it didn't include delete in the first place? I don't know that I can defend my idea of having a seperate delete own pages permission against a standard of consistency. If all the other edit own permissions include the delete pages permission, then I have no quibbles. :) Except it would be more userfriendly if I knew what the permissions meant before I granted them to someone, without having to look at the code. But I do realize that is completely out of the scope of the patch. :) Where do all these little changes *go*, anyway? The changelog just says things like... refractored the search module. Anisa. ------------------------------------------------------------------------ Sat, 30 Apr 2005 16:04:14 +0000 : clydefrog I believe that "maintain books" permission implies the ability to delete book pages, so this patch fixes that bug. Other node modules include deletion in the "edit own" permission. For example, page.module: <?php if ($op == 'update' || $op == 'delete') { if (user_access('edit own pages') && ($user->uid == $node->uid)) { return TRUE; } } ?> I agree that it would be better if permissions were clearer about what they implied. There was a discussion started by killes [1] a while ago: "One problem with Drupal is the way it handles user permissions. The permissions are attached to the user object but nobody really knows what they do. To find out what a particular permission allows you to do, you often need to have a look at the code. " I don't know if it ever lead to any code. I think the changelog is meant to list the changes since the previous version without too much technical detail. If you want to see every last change, take a look at the cvs messages [2]. [1] http://lists.drupal.org/archives/drupal-devel/2005-04/msg00704.html [2] http://drupal.org/project/cvs/3060 ------------------------------------------------------------------------ Mon, 23 May 2005 17:33:18 +0000 : rivena I've been looking at the book module permissions recently, and did a bit of testing. I wanted to know what was the minimum permissions needed to create a new book. I assumed based on this discussion, and a similar discussion on the drupal support list that this was 'maintain books' and 'administer nodes'. I ran into a problem, then added 'access nodes' to this. So now my user has no other permissions other than 'maintain books', 'administer nodes' and 'access nodes'. You'd think this is more than sufficient to do something simple like create a new book, but in fact, with just these permissions, you do not have access to create a book page. You can edit other pages, and move them into a new book. Once you have create pages permission, you appear to finally have all the permissions needed, edit own pages is not needed. So, since, the point of this patch was that not having delete permission was odd, might I suggest adding create permission to maintain books? ;p Anisa. ------------------------------------------------------------------------ Mon, 08 Aug 2005 10:37:49 +0000 : Bèr Kessels I would say this is a "by design". Maybe we need to rename the two book permissions into: "create books" and "edit own book pages". None of the other nodes can be deleted by the roles with "edit own Foo" permissions, nor should the books. ------------------------------------------------------------------------ Mon, 08 Aug 2005 16:59:30 +0000 : clydefrog Thanks for your review, Ber. story.module [3] allows users with "edit own stories" permission to delete stories. <?php if ($op == 'update' || $op == 'delete') { if (user_access('edit own stories') && ($user->uid == $node->uid)) { return TRUE; } } ?> Therefore it is not inconsistent for the same behavior to apply to book nodes. If "maintain books" really means "create books", then the name should be changed and existing code in book.module needs to be updated. However, I think "maintain books" should allow more than just that. In my situation, I want to be able to give someone the permissions to create books and to create, edit, and delete all book pages. With my patch, "maintain books" and "create book pages" does that. [3] http://cvs.drupal.org/viewcvs/drupal/drupal/modules/story.module?rev=1.167&… ------------------------------------------------------------------------ Mon, 08 Aug 2005 19:16:35 +0000 : Bèr Kessels What i see from story, that is specific to story.module too. We should really settle for a consistent behaviour for *all* nodes, one that is "enforced" from node.module. These things should really not be decided per node-type and per-module. Thats a recipe for a mess. ------------------------------------------------------------------------ Mon, 08 Aug 2005 19:25:27 +0000 : Eric Scouten I agree that the current system is a mess. I disagree that there should be one consistent policy enforced by node.module. Sites may have quite legitimate reasons to have different policies for different node types. ------------------------------------------------------------------------ Mon, 08 Aug 2005 20:35:59 +0000 : Bèr Kessels What i was aiming at is to let node module take care of the following: * edit $node_type * administer $node_type * delete $node_type (but would that not better be in 'administer';: users who can administer can delete) * create $node_type That way the modules need not cre about this. ut they /can/ take care of it, if they wish, and add more rules and permissions. ------------------------------------------------------------------------ Mon, 08 Aug 2005 20:36:32 +0000 : Bèr Kessels oh, and -forgot- : * view $node_type

1 0