[DRE-maint] Bug#907433: ruby-ffaker: FTBFS (TestLoremFR fails)

[Santiago Vila]

reopen 907433
retitle 907433 ruby-ffaker: FTBFS randomly (TestLoremFR fails)
thanks

On Wed, Feb 05, 2020 at 11:33:47AM +0100, Pirate Praveen wrote:
> On Mon, 27 Aug 2018 22:38:21 +0000 Santiago Vila <sanvila at debian.org> wrote:
> > I tried to build this package in buster but it failed:
> 
>  I was able to build it without any failures (logs attached). Please
> reopen if it is still an issue.

Yes, I believe it is still an issue.

The failure really happens randomly (not always, as it could be
inferred from the old bug title), so I'm using retitle to fix this.

The build history at reproducible-builds confirms that the package has
been failing randomly for a long time:

https://tests.reproducible-builds.org/debian/rb-pkg/unstable/amd64/ruby-ffaker.html

Please have a look at the failure itself: I don't know a lot about ruby,
but the test is apparently trying to use a string which does not contain
the letter À to match the string "À maintes prémâchées lui." as shown
by the build log:

</\A[ -.àâéèêîôùûa-z]+\z/i> was expected to be =~ <"À maintes prémâchées lui.">.

It seems that the test selects a phrase randomly from a large corpus,
and this explains the random failures.

I can think of two ways to fix the FTBFS problem:

The first one would be to fix the test so that it does the right
thing, which I agree it could take time and effort.

The second one is to disable the test, since (IMHO) it is clearly wrongly
designed.

In cases like this (when I'm aware that the failure happens randomly
with low probability, which I was not when I initially reported this),
I usually report the FTBFS as "important", on the basis that trying a
few more times the package will eventually build. According to that,
please free to downgrade the severity, but in this case I believe it
is quite obvious that the test is wrongly designed and should be
disabled. Packages are expected to build all the time, not just "most"
of the time.

Thanks.