Bug 26666 – bugzilla-tool continues execution even if subcommands fail

• Status: RESOLVED
• Resolution: FIXED
• Component: Tools / Tests

[Eric Seidel (no email)]

bugzilla-tool continues execution even if subcommands fail

I've noticed in a couple places that bugzilla-tool will continue executing even if the underlying "git svn rebase" command fails (due to a conflict for instance).

This is bad, and we should make bugzilla-tool more robust against this kind of failure by throwing ScriptExceptions (that's what there designed for).

That will allow the eventual land-everything-in-the-commit-queue mode to work and recover from build/update failures, while still allowing normal usage to error out correctly.

[Eric Seidel (no email)]

Created attachment 31760 [details]
Make SCM.run_command smarter and use it more


---
 2 files changed, 47 insertions(+), 49 deletions(-)

[Dimitri Glazkov (Google)]

What the heck kind of bug number is that?

[Eric Seidel (no email)]

Clearly bugzilla-tool ate from the tree of knowledge and must now be banished from the garden of goodness.

[David Levin]

Comment on attachment 31760 [details]
Make SCM.run_command smarter and use it more

Only one comment:

> diff --git a/WebKitTools/Scripts/modules/scm.py b/WebKitTools/Scripts/modules/scm.py
> +    def run_command(command, cwd=None, input=None, raise_on_failure=True, return_exit_code=False):
> +        stdin = subprocess.PIPE if input else None
> +        process = subprocess.Popen(command, stdout=subprocess.PIPE, stdin=stdin, shell=True, cwd=cwd)
> +        output = process.communicate(input)[0].rstrip()
> +        exit_code = process.wait()
> +        if raise_on_failure and exit_code != 0:

Use:
        if raise_on_failure and exit_code:

[Eric Seidel (no email)]

Committing to http://svn.webkit.org/repository/webkit/trunk ...
	M	WebKitTools/ChangeLog
	M	WebKitTools/Scripts/bugzilla-tool
	M	WebKitTools/Scripts/modules/scm.py
Committed r45026
http://trac.webkit.org/changeset/45026