1209772 - 'mozregression was installed. please re-run your command.' when running ./mach mozregression

Status: RESOLVED FIXED

(Firefox Build System :: Mach Core, enhancement)

Description

[Benoit Girard (:BenWa)]

Run log:

~/mozilla/mozilla-central/tree> ./mach mozregression --help
mozregression was installed. please re-run your command.

Re-running the command gives the same result. I've already installed mozregression outside of mach in case that makes a difference.

Comment 1

[Julien Pagès (:parkouss)]

Sorry for the late response here. So, I believe this is because you have a version of mozregression outside of mach that is inferior to 1.0.0. In that case, updating it one time using "pip install -U mozregression" should fix the issue.

Does that works ?

Comment 2

[Benoit Girard (:BenWa)]

It's fixed. Ideally we should warn but WONTFIX is probably alright here. It shouldn't be had to add an unconditional:

mozregression was installed. please re-run your command.
If you keep getting this message please run: pip install -U mozregression

Comment 3

[Julien Pagès (:parkouss)]

Agreed, looks like a good idea to me. There are minor chances that people have a system wide "old" mozregression, but when that happen, the current message is not cool. :)

Comment 4

[Julien Pagès (:parkouss)]

Attachment: MozReview Request: Bug 1209772 - 'mozregression was installed. please re-run your command.' when running ./mach mozregression. r=ahal

Bug 1209772 - 'mozregression was installed. please re-run your command.' when running ./mach mozregression. r=ahal

Comment 5

[Andrew Halberstadt [:ahal]]

Comment on attachment 8669905 [details]
MozReview Request: Bug 1209772 - 'mozregression was installed. please re-run your command.' when running ./mach mozregression. r=ahal

https://reviewboard.mozilla.org/r/21305/#review19187

This is an improvement, so r+.

But this will happen every time a new mozregression is released to pypi. What if we first tried to import and update mozregression globally and then only activate the virtualenv if that fails? If that's hard for any reason, then just ignore this comment. It's a pretty minor problem and not worth a lot of time.

Comment 6

[Julien Pagès (:parkouss)]

(In reply to Andrew Halberstadt [:ahal] from comment #5)

> But this will happen every time a new mozregression is released to pypi.
> What if we first tried to import and update mozregression globally and then
> only activate the virtualenv if that fails? If that's hard for any reason,
> then just ignore this comment. It's a pretty minor problem and not worth a
> lot of time.

No, I should have explained more - this will only happen if the mozregression version is < 1.0.0, because there was no mach interface in mozregression before that.

I mean I added a new mach_interface.py file in mozregression, and that is the entry point for mach. So if you look in the code, it try to first import mozregression.mach_interface - if that fails, it calls

> cmd.virtualenv_manager.install_pip_package('mozregression')

which I suspect don't try to update. So nothing changed, it will do that over and over. Well I suspect to really fix that I can use
the "_run_pip" method (as I did for the upgrade part, since there is no public interface for that in virtualenvmanager):

> cmd.virtualenv_manager._run_pip(['install', '-U', 'mozregression'])

Now that I look at it, this fix would be better. Andrew, what do you think ?

Comment 7

[Julien Pagès (:parkouss)]

Ok, so we chatted about it on irc, and decided to land this patch like this.

Comment 8

[Pulsebot]

https://hg.mozilla.org/integration/mozilla-inbound/rev/7fb566dcac63

Comment 9

[Wes Kocher (:KWierso) (Not reading bugmail; email directly if needed)]

https://hg.mozilla.org/mozilla-central/rev/7fb566dcac63