Bug 81411 – Add a land-from-url command

• Status: RESOLVED
• Resolution: FIXED
• Component: New Bugs

[Eric Seidel (no email)]

Add a land-from-url command

[Eric Seidel (no email)]

Created attachment 132387 [details]
Patch

[Adam Barth]

Comment on attachment 132387 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=132387&action=review

> Tools/Scripts/webkitpy/tool/commands/download.py:213
> +            bug_id = urls.parse_bug_id(url)
> +            if bug_id:
> +                patches = tool.bugs.fetch_bug(bug_id).patches()
> +                log("%s found on bug %s." % (pluralize("patch", len(patches)), bug_id))
> +                all_patches += patches

It's a bit odd that this works differently than land-from-bug.  In person you gave me a big speech about how land-from-bug is confusing and wrong, so I guess that means you plan to fix land-from-bug?  ;)

[WebKit Review Bot]

Comment on attachment 132387 [details]
Patch

Clearing flags on attachment: 132387

Committed r111095: <http://trac.webkit.org/changeset/111095>

[WebKit Review Bot]

All reviewed patches have been landed.  Closing bug.

[Dirk Pranke]

For those of us not there in person ... what is the point of this command? "land all the patches at a given URL" sounds dodgy on the face of it.

[Adam Barth]

(In reply to comment #5)
> For those of us not there in person ... what is the point of this command? "land all the patches at a given URL" sounds dodgy on the face of it.

Any more dodgy than the land-from-bug command?  We want to be able to accept patches from various sources besides just bugs and attachments (e.g., libpaste, GitHub).  A URL is a nice universal designator.

The plan is to whitelist the sources, at least at first.  Notice that this command only currently accepts bugs.webkit.org URLs.

[Dirk Pranke]

(In reply to comment #6)
> (In reply to comment #5)
> > For those of us not there in person ... what is the point of this command? "land all the patches at a given URL" sounds dodgy on the face of it.
> 
> Any more dodgy than the land-from-bug command? 

Sure; land-from-bug at least sounds like you're poking at a structured, approved place.

At any rate, I have no objection to landing patches retrieved from arbitrary URLs (since presumably you know that the url is a patch), but I wasn't sure if you were planning to scrape pages looking for embedded links to patches, or what, that's all. Just looking to understand what you have in mind ...

[Adam Barth]

We also have a dream of having sheriffbot land patches (e.g., specified by URL).  There's some trickiness around authentication, of course.

[Dirk Pranke]

(In reply to comment #8)
> We also have a dream of having sheriffbot land patches (e.g., specified by URL).

When would you want to do this and not just use the CQ?

[Adam Barth]

> When would you want to do this and not just use the CQ?

For example, if you have rollout that you want to land right now (and don't want to wait for the CQ), or if you have a patch that just changes the web site and you don't want to wait for building/running the tests.