Re: Emacs-diffstrunk r117208: * lisp/files.el (locate-dominating-file): Expand file argument.

[Glenn Morris]

Michael Albinus wrote:

>> BTW, I think that test should set the BZR_HOME environment variable,
>> like test/automated/vc-bzr.el does. This will be good not just for
>> hydra, but to stop tests writing to ~/.bzr.log in general.
>
> Yes. However, it's not guaranteed that tramp-test29-vc-registered
> chooses bzr. It could also take git or hg, whatever it finds on the test
> machine.

It will be harmless to set BZR_HOME if something other than bzr is used.
But you should be able to make the setting conditional on bzr being
chosen, if you like. (I could not figure out if I needed to modify
process-environment or tramp-remote-process-environment, and if the
latter needed to be changed early on.)

> A bzr only change will be good for hydra, but maybe we need some more
> general consideration here.

As I said, I think it will be good for everyone to set BZR_HOME to a
temp value.

> OTOH, it's not guaranteed that all people have set BZR_HOME.

I don't understand.

> Couldn't this setting hide problems, which would be detected otherwise
> in the tests?

It only hides the problem of a non-functioning bzr due to a missing HOME
directory, which is not Emacs's problem IMO.