Page MenuHomePhabricator

Allow downloading a revision 'commit', meaning including the metadatas
Closed, DuplicatePublic

Description

Currently there is a "Download raw diff" button, but that does not include metadatas (in git for example the usual way of providing diffs is to do git format-patch, which will create diffs containing metadatas such as the commit message, the author, the commit hash etc..) and we should be able to retrive that exact same commit, including the netadatas, first through the web interface, and secondly using the REST conduit API.

Event Timeline

thiblahute renamed this task from Allo downloading a revision 'commit', meaning including the metadatas to Allow downloading a revision 'commit', meaning including the metadatas.
thiblahute raised the priority of this task from to Needs Triage.
thiblahute updated the task description. (Show Details)
thiblahute added a project: Differential.
thiblahute added a subscriber: thiblahute.

use arc patch for that.

This answer is just not satisfatory, we should not be requiring people to install arc + php + ... to just try a patch out. They should be able to just get the patch do a git am diff.path and try it, rebasing if needed.. etc.

Specifically, if they are not willing to use arc, then it's probably not important for them to retain meta-data anyway.

To be able to know on what commit the patch was base and most importantly rebase it, you need to have the metadatas. What is your concern about adding that feature?

I'm not dead-set against it (And it's not my call anyway). but:

  • You didn't read Contributing Feature Requests. If you had, you'd provided a better task description.
  • This ticket basically boils down to "I/users don't like arc/phabricator", which is not a fun ticket to deal with. The actual need ("apply pending change locally") is already well-supported, in multiple ways, just not in the way you like.
  • "Don't like arc" is already covered by T5000.
  • Phabricator is not limited to git. What about hg? what about svn? what if the author uses svn and the reviewer uses git?

I use git phab and am very happy with it, it uses arc in the background (at least for now).

Problem is not about me but about other contributors who think this issue is a blocker for the adoption of phabricator in the GStreamer community, so I would rather move forward and find a solution.

Basically what I am saying here is that users should be able to get, through the web interface, exactlt what the user first propsed, which in the case of git is a git commit, and for other vcs, something else.

  • This ticket basically boils down to "I/users don't like arc/phabricator", which is not a fun ticket to deal with. The actual need ("apply pending change locally") is already well-supported, in multiple ways, just not in the way you like.

The actual need is to have a very straight forward way to let our contributors test a patch without having to install anything, testing a patch might imply rebase it etc. I as a user like very much phabricator and want to see it adopted by the community ;)

chad added a subscriber: chad.

This is covered well enough in the description of T5000:

In fact, there is a way to bypass Arc, but you're expected to 'git diff' and copy/paste that in the web UI. This is really inconvenient. And then you'll loose commit metadata that way. Like committer, author, and most important: parent of a commit.

T5000 cover the "Let users play without arc".
T1508 is related: the git commits the make up a revision are not uploaded ATM. They can be reconstructed, which is what arc patch is doing. See also https://secure.phabricator.com/book/phabricator/article/harbormaster/#change-handoff for an advanced, git-only approach for this.

OK, to me those are 2 very different issues, T5000 is about having a workflow integrated into git and this one is just about allowing users to download the diffs submited to phab in the same exact same form as proposed instead of a stripped down version... I am not sure how you can say they are the same issues, but anyway, looks like there is no way we agree on that at that point (second time you close the issue, in one day...)

Thanks for being so responsive.