Robert Douglass wrote:
In the sandbox discussion, an interesting pattern has appeared. Virtually every developer who is involved with Drupal.org infrastructure at a personal level (those named in the title) want to bring order to
The actual proposal was actually brought up by somebody else and was to completely remove the sandboxes. I opposed this and hence see it as a bit ironic that you mention me first...
the sandboxes and crowbar the "not projects" that live there into the new release system. Everybody else is against this. Worse, we don't seem to be moving towards agreement.
I am not that pessimistic. Most people seem to understand now that module development should be in /modules.
Usually in a case like this, one side of the story isn't fully understood. I have to admit that I still don't understand what is so bad about the sandboxes that makes the people involved with infrastructure hate them. I use my sandbox only infrequently but am very glad it is there.
Why? A recent commti you did was your memcached code. Does this need to be in a sandbox? Shouldn't this rather be an attachment to an issue or even a project?
The answer can't be "the code there is lousy" because that applies to at least 200 of the projects in contrib as well (a number which will rise with the new proposed changes). So what is it that makes the sandboxes a thorn in the side of the people who actually have to maintain the infrastructure?
Kjartan has explained that quite well. As far as I see it there are the following use cases for sandboxes that don't fit the current rules: - share some code snippet (Karoly, Goba) - develop patches for contrib modules Of these only the last one needs (IMO) revision control and we could amend the rules to allow for this. For the first use cases a file upload to an issue or the snippet repository in the handbook should be fine. It is actually better there because it can be more easily found.
The work that goes into Drupal.org infrastructure is chronically under-appreciated, and if the sandboxes are a special burden that I (we) haven't recognized, it would help the conversation to find that out.
They are simply uncontrollable, nobody is really accountable for the stuff in there etc. And most of the stuff is crap that nobody will understand, ie is completely undocumented and thus unusable by anybody but the author. Example: http://cvs.drupal.org/viewcvs/drupal/contributions/sandbox/yched/ I initially was delighted to find this, because there is some nodereference issue I need to get resolved. But what am I supposed to do with this? "Fisrt test commit :-)" is not a useful commit message and the README is only the original. Most of the other sandboxes are similar: Undocumented, non-working, age-old stuff. 98% inside these sandboxes is crap not worth keeping. Cheers, Gerhard