checkbox-dev team mailing list archive

[Ara Pulido <ara.pulido@xxxxxxxxxxxxx>]

On 11/06/14 17:44, Zygmunt Krynicki wrote:
> Sure.
> 
> The default behavior is what we currently do, strip all localization
> settings inherited from the tester's environment.
> 
> Let's say you are running a test that shows a window and tells you how
> to proceed in that window. Even if we do everything required to get that
> localized, translate every single message to every language on the
> planet, all of that will never be used. PlainBox disables localization
> (by pretending to run on "80s POSIX system) for all external programs it
> launches.
> 
> So as we go and fix our scripts we don't need that behavior. Instead, if
> a script is localized (like all scripts should be) it will use the
> natural language that the tester expects to see. If a particular script
> needs to run an external command and parse the output it can handle
> disabling l10n internally, just for that command. Here's where the flag
> comes in handy, as we go and fix each script (one by one) we can add the
> 'preserve-locale' flag to jobs that run that script. Eventually all our
> scripts will use that flag and the entire user experience will be
> localized, limited only by translation coverage.

It does, thanks a lot.

> 
> I hope that explains it.
> 
> Best regards
> ZK
> 
> 
> 
> On Wed, Jun 11, 2014 at 4:52 PM, Ara Pulido <ara.pulido@xxxxxxxxxxxxx
> <mailto:ara.pulido@xxxxxxxxxxxxx>> wrote:
> 
> 
> 
>     On 11/06/14 14:59, Zygmunt Krynicki wrote:
>     > Hey
>     >
>     > As I was reviewing a branch today I realized that we cannot have any
>     > localized tests at this time. This is because we purposefully disable
>     > l10n at plainbox level out of concern for predictable behavior for
>     > things like error messages and parsing.
>     >
>     > I've filed a bug https://bugs.launchpad.net/plainbox/+bug/1328903 that
>     > offers a way out, that allows us to continuously improve the situation
>     > and enable subsequent jobs to have fully localized testing user
>     > interface while not having any regressions or flag days.
> 
>     I have tried to understand this email and the bug report, but I haven't
>     been able :)
> 
>     Can you put an example on why this flag is needed and what would be the
>     default behaviour?
> 
>     Thanks,
>     Ara.
> 
>     >
>     > I'd like your feedback on the idea. Given the absence of feedback,
>     I'll
>     > just implement it.
>     >
>     > Thanks
>     > ZK
>     >
>     >
> 
>     --
>     Mailing list: https://launchpad.net/~checkbox-dev
>     Post to     : checkbox-dev@xxxxxxxxxxxxxxxxxxx
>     <mailto:checkbox-dev@xxxxxxxxxxxxxxxxxxx>
>     Unsubscribe : https://launchpad.net/~checkbox-dev
>     More help   : https://help.launchpad.net/ListHelp
> 
>