Re: ESC call Thursday 16:00 central European time ...

On Wed, 2014-10-08 at 15:06 +0200, Jean-Baptiste Faure wrote:
> Hi,
>
> Le 08/10/2014 14:10, Bjoern Michaelsen a écrit :
> > OTOH a _much_ better point for this discussion would had been when the original
> > patch was still on review. 
> The problem is not the original patch (the patch for the master), it
> works perfectly and Kohei did a fantastic job with it and its
> complementary patch which added the configuration option allowing the
> choose between the new behavior and the former. Thank you very much for
> that.
>
> The issue is in the decision to backport the first patch without the
> configuration option (fdo#81309 only) to 4.3.1 and 4.2.7

It was because of the translation requirement that the backport would
add.

>  as if this
> former behavior was completely broken.

There were those users who believed that the former behavior was
"broken", and fought tooth and nail to get that "fixed".

>  Ok, it is not perfect but many
> spreadsheets are build on this behavior since many years. 

Which I was not even remotely aware of, but then I'm not aware of 100%
of how our users use our software.  I get surprised every single day.

> With 4.3.2 and
> 4.2.7 all these spreadsheets do not work anymore 

By the same token, the "fix" also allows those users wanting the fix to
start using Calc to fit their needs.

> and the users have to
> heavily modify the logic of their spreadsheets if they want to continue
> to use sort with references.

Sure, any changes are evil.

>  Please try the test file in bug 81633
> yourself. Compare sorting in 4.2.6 and in 4.2.7 or the master (with the
> new behavior).

And please do start talking with those users who asked for that fix in
the first place.  All you are doing right now is shooting the messenger.
And I can tell you it's a terrible place to be in.

> For me (and I am not alone to think that, see the growing number of
> duplicates of bug 81633), such big change should be done only for a
> x.y.0 version with a clear information in the release notes.

To be fair, there were also a fair number of duplicates for the "bug" I
"fixed" that you disapprove of.

Please make a decision for us then, and work on putting what you think
is the correct fix in.  I think that's fair.

That's all I have to say on this issue.  I'll stay out of this from this
point on.

Kohei