>> > This "breaks unrelated software" only if you have configured that
>> > software
>> > to look at the contents of this package.  Do you intend to also claim
>> > ca-certificates "breaks unrelated software" every time it drops a CA
>> > certificate because they determine the certificate authority isn't
>> > trustworthy?
> 
>> This is not the case here ! There is only a renaming.
> 
> You didn't answer my question.

The implied answer was "no".

>> The fix is easy and this bug should stay critical to not slip out of  Etch.
>> If a function was renamed in libc, would you say that it breaks unrelated
>> software only if you have configured that software to use this function ?
> 
> No, because it wouldn't break any unrelated software; it would only break
> *related* software.  That would be treated as a serious bug -- the same as
> a bug where libc dropped a function.

OK, I get the point.

> And it would be treated such because library functions are something we
> guarantee.  Interfaces to particular CA certificates, however, are not
> something we as a project guarantee.  I understand that it's an
> inconvenient
> upgrade problem for users who link to this certificate, but that doesn't
> make it 'critical'.

Well, I hope that the fix will go into Etch since it will permit smooth upgrade
from people using ca-certificates from bpo and relying on CAcert certificate.


Reply via email to