Recent blog entries for mitch

Ok, today I can't resist to contribute to the cert level/metric discussion.

I just came across a new Lead Developer of The GIMP and I suddenly felt like I had to post a feature request.

Not that I've certified any idiot, but just in case it happens one day, I'd like to uncertify this person.

I totally agree with kuro5hin's response to the Outcast idea ("not wanting to be a member of a community that defines itself by exclusion rather than inclusion"), but I want to simply ignore idiots by not certifying them at all, even if I have certified them in the first place.

Just my 2 Pfennig.

Got Gimp installed on my university's Solaris boxes and was shocked by the size of the resulting executable. With debugging info it's 33MB (!!!). Well, but I want debugging info so users can send me stack traces if it crashes...
Guess I'll just not strip it and wait until someone comes to kill me for causing 33 MB going over NFS all the time :)

The 8 bit displays of the Solaris machines made me starting Gimp in indexed mode for the first time, and voila, broken.
All Gimp plug-ins which have previews share the same piece of code to get their colormap right. And this sequence is

  • deprecated because it doesn't use GdkRGB.
  • missing in all plug-ins which don't have a preview, causing bad flickering on the display.
Which means I have to add another libgimpui function to do it in a consistent manner for all plug-ins which have a ui...

Which means I have to hack all of them again, oh no!

It seems I'm the only person on the northern hemisphere who didn't see the polar lights which are told to be there for two days now. Even worse, after a trip to the countryside where it's dark enough at night, I'm now back in Berlin where the "light pollution" (??) is too massive to see it, sigh.

Anyway, back to GIMP hacking. And there comes the next oddity: I one again browsed all bugs and wasn't able to reproduce most of them. Either because they occur on some esoteric hardware or they simply don't occur on my hardware.

Well, this wouldn't be too bad if the bugs I can't reproduce didn't sound like stuff that prevent a fast realease (like SEGVs and FPEs in commonly used core functions).

Last but not least my reorganisation of the display shell re-broke "Shrink Wrap" and the shrink_wrap code looks like being designed around a dozend window manager bugs. needless to say that shrink_wrap works fine here, too.

Which leads me to the conclusion that it was a bad idea to configure my system to behave nicely. What I need is a broken X server, a broken window manager, a broken include and library structure and a non-broken hardware (since most people out there - including me - seem to hack on "broken" Intel machines).

But you didn't hear the worst thing yet: Today is a sunny day and I will not fix any bug but go out and enjoy the sun :-)

Have a nice sunday!

Finally back home... GUADEC was so cool.

I was really happy to meet all the people I knew by mail only, esp. all the GIMPers. Had a nice time with Sven, Tigert, Raphael, Nether, ...

Watched Tigert making two new GIMP icons in 5 minutes, amazing.

Before going back to Berlin, I first went to Sven's place where we managed to make GIMP un-compileable at 4 o'clock in the morning :-). Well, it seems all the people who are updating daily were at GUADEC, because we got no complaints.

After a hacking-free day yesterday, busy mode has returned and I had to answer students' questions in a test this morning. This afternoon we are going to correct it. Will not going to be fun, as there were 800+ students writing the test, eek. This should keep me busy for the next two days.

Got a bit crazy about correct context sensitive mouse cursors in GIMP's tools during the last two days.

After adding "Zoom in" and "Zoom out" cursors to the magnify tool (which is not really a bugfix), I found that the "Intelligent Scissors" were displaying totally nonsense cursors and fixed the tool (and yes, this is a bugfix :-).

The remaining problem is that many tools still need a mouse motion before they update their cursors. Immediate cursor updates however give much better feedback (it simply feels natural, i.e. "how expected", if the cursor changes as soon as e.g. a modifier key was pressed).

I think I'll look at the other tools as well...

It seems that my GIMP plug-in hacking has caused the print plug-in (which is mainly developed at sourceforge) to be totally incompatible with GIMP 1.0 and even all GIMP versions before 1.1.17

But it also seems that this problem seems to be solved soon, as it pointed both sides to the fact that we need some sort of core/ui separation for this plug-in. Doing this will also help turning the plug-in into a standalone application which can be plugged into a variety of applications (like sketch).

Synergy at work?

Anyway, if you've seen the output of the Gimp-Print plug-in, you'll love to have it pluggable into other apps. Believe me :-)

[entry updated]

Got memprof installed last night -- immediately pointed me to several huge GIMP leaks.
What a cool tool (TM).

Go and get it from Owen's page!

New Advogato Features

New HTML Parser: The long-awaited libxml2 based HTML parser code is live. It needs further work but already handles most markup better than the original parser.

Keep up with the latest Advogato features by reading the Advogato status blog.

If you're a C programmer with some spare time, take a look at the mod_virgule project page and help us with one of the tasks on the ToDo list!