Re: [PATCH] Gracefully handle incompatible locale data

[Ludovic Courtès]

Allan McRae <address@hidden> skribis:

> On 29/09/15 06:54, Carlos O'Donell wrote:
>> On 09/26/2015 06:24 AM, Ludovic Courtès wrote:
>>> Furthermore, the function in question returns EINVAL in other similar
>>> cases–e.g., when libc 2.22 loads LC_COLLATE data from libc 2.21.
>> 
>> If you change this particular case to EINVAL, what does the user see
>> as a result of this change? Do they get a non-zero exit code from
>> `localedef --list-archive` along with an error written out to stderr?
>> 
>> This is the kind of change I'm expecting. If we are removing an assertion,
>> we should be replacing it with something meaningful and verifying that
>> meaningful change.
>> 
>> You need not change any of the other cases you've found that return EINVAL,
>> we can update those incrementally, but for this one change you're making
>> we should fix it as best we can.
>> 
>
> If I am reading this correctly, the change to from an abort to EINVAL
> would be fine if it is accompanied by a change to localedef
> --list-archive.  Is that correct?

My understanding is that no such change is needed, but I’m waiting for
confirmation or clarification:

  https://sourceware.org/ml/libc-alpha/2015-09/msg00727.html

> A solution to this would be great given we now run into this assert with
> locale archives built with different glibc builds along the 2.22 release
> branch.

I’m glad you value the practical benefits.  ;-)

Ludo’.