[Ohrrpgce] SVN: james/3927 Backport the fix for text box instead-chaining bug 824 to the ypsiliform

[James Paige]

On Mon, Oct 25, 2010 at 06:14:03PM +1300, Ralph Versteegen wrote:
> On 25 October 2010 10:46, James Paige <Bob at hamsterrepublic.com> wrote:
> > On Mon, Oct 25, 2010 at 10:17:12AM +1300, Ralph Versteegen wrote:
> >> On 24 October 2010 13:00, James Paige <Bob at hamsterrepublic.com> wrote:
> >> > On Sat, Oct 23, 2010 at 04:58:22PM -0700, subversion at HamsterRepublic.com wrote:
> >> >> james
> >> >> 2010-10-23 16:58:22 -0700 (Sat, 23 Oct 2010)
> >> >> 82
> >> >> Backport the fix for text box instead-chaining bug 824 to the ypsiliform+3 branch
> >> >> ---
> >> >> U   rel/ypsiliform/whatsnew.txt
> >> >> U   rel/ypsiliform/yetmore.bas
> >> >> U   rel/ypsiliform/yetmore.bi
> >> >> U   wip/whatsnew.txt
> >> >
> >> > Can anybody think of a reason not to release ypsiliform+3 right now?
> >> >
> >> > There are some good bugfixes in there, and we can always do ypsiliform+4
> >> > when we resolve the ogg and mp3 import/convert bugs.
> >> >
> >> > ---
> >> > James Paige
> >>
> >> Wow, I'd thought we gave up on Ypsiliform+3.
> >> There are lots of other bugfixes in wip, so shouldn't they be
> >> backported too? It's a bit cruel to withhold them from people isn't
> >> it?
> >> But instead of doing that, why not just release Zenzizenic soon? It's
> >> been a year since the last major release!
> >
> > I like that even better :)
> 
> OK... so how soon? I had better delay the textslice/textbox rewrite
> until afterwards, but I could

I can make a release candidate this afternoon. I think the nightlies are 
pretty stable right now.

Even bug 826 is in my mind just a nice-to-have fix rather than a 
must-have-for-release because releasing with that bug unfixed wouldn't 
be any worse than letting people continue to use ypsiliform.

But if we wait on any bug, I guess it would be that. Nothing else looks 
dire to me.

> As explained on bugzilla there are two bugs: the one which causes
> conversion to fail, which is just a filename quoting problem, and the
> one which sometimes causes gfx_sdl to lose all sense of the keyboard
> state (especially when conversion fails?), which is somehow caused by
> opening madplay/oggenc in a foreground terminal window with SHELL. I
> think that one's fixable by just using alternative platform specific
> spawning commands instead of SHELL. SHELL does some crazy stuff, and I
> noticed that the Unix implementation is also buggy: it leaks file
> descriptors, and it looks like a giant hack which I would rather
> avoid.

Hmmm... I was checking out the filename quoting problem, and it seems 
correct already. Maybe the quoting is being messed up by some crazyness 
of SHELL?

---
James