Re: 1.3.15 reminder / must-fix bugs

[Havoc Pennington <hp redhat com>]

Hi,

I'm here for a couple hours still.

Owen Taylor <otaylor redhat com> writes:    
> > I had #59101 but the milestone has been moved to 2.0.1, which means, I guess,
> > that this bug is no longer appearing very often, in which case the 2.0.1
> > milestone sounds fine.
> 
> It looks to me like Havoc is saying "this is fixed, but I'm not sure there
> isn't something wrong internally."

Yes, there is a small internal bug that could cause a display glitch
in some obscure cases. No one is going to notice, most likely. 
Should be fixed eventually but not must-fix.

> > #67999 is definitely user unfriendly and makes the new text widget almost
> > unusable for any serious usage, which would be too bad for such a powerful
> > widget.
> 
> Well, I'm not sure I'd say "unusable". It's basically comsmetic. But
> yes, it will look bad if this is left around. I'd agree this is is a
> mustfix.

A high-priority issue.

> > >  - Would cause compatibility problems if fixed later.
> > 
> > #58742 is in this category.
> 
> We can always add another entry point. We aren't making incompatible
> API changes at this point, at least. What I mean by compatibility
> problems are things like installation locations, RC file syntax, and
> stuff that would be very hard to do while maintaining ABI
> compatibility.
> 
> (And yes, strlen is darn fast compared to most of the stuff the text
> widget does...)

My opinion on this is that if you are making a copy of a huge hunk of
text, that's a much bigger deal than the strlen(). You can always make
your algorithms work natively on the buffer instead of doing a
get_text(). However I'm open to adding
gtk_text_buffer_get_text_with_len() in 2.2 if this is really showing
up in profiles.

Havoc