Well, NetHunter won't be able to apply my fixes until Saturday, so the UI changes will hafta wait. However, I did add 4GB file support and got to clean up the interfaces (in a API sense, not a UI sense) a bit. Hopefully I didn't break anything after all the stress testing I did on the code a few days ago...

Oh, a word on the 4 gig file support: it seems to be entirely supported by the ICQ protocol, but it requries an OS can support files larger than 2GB on both ends. I was thinking of adding real large file support (i.e. 64 bits), but that would require extending the ICQ protocol, and using 64bit file pointers and size variables everywhere, and its really not worth it right now. GnomeICU can currently deal with 64bit files internally well enough, but it won't be able to express sizes larger than 4GB to the other end or calculate percentage/KBPS once you cross the 4GB boundry. So, yeah, being able to express a whole 2 more gigs is a pretty good deal, but don't expect anything better any time soon...

Totally hacked apart the file transfer for the GnomeICU 1.0 bugfix cycle. Compared to the code in CVS, my tree is much more stable, and gets significantly better throughput (~25% on a cable modem), without adding any new features. It's also rather less glitchy...

I'll send NetHunter the patch tommorow, and then start the file transfer UI overhaul for NetHunter's planned UI fixup release. Will probably also add a few small other features, but the file transfer code is virtually feature complete as it is.

But I gotta sleep right now, it hurts to live...

I've stabilized batch file transfers quite nicely, and fixed the inefficencies in receiving files left over from the original GnomeICU code. I've also done alot of file transfer UI hacking recently, I'm about three quarters done the UI changes targeted for the next patch. The file transfer code is very near release quality, now all I need is free time to finish it off and send it away to David.

I've decided that auto accepting of file transfers will make it in to this patch, because the UI changes worked out just right, so receiving file transfers and auto accepting file transfers share about 85% of their code in a way which is not immediately obvious, and the saving/loading of the auto accepting configuration is already in the CVS. So, it will be almost trivial to add the few extra lines necessary to activate auto accepting.

More GnomeICU hacking, got multi-file transfers working correctly, now for some horrible UI hacking before sending a release candidate patch off. Might implement auto-accpeting of file transfers in this patch to, but it's more likely to get thrown in with patch #3 - "finishing touches/brown paper bag fixes"

I''m terribly sleep deprived and feeling a bit off, so I think I'll sign off for tonight, but first, my final thought: I NEED MORE OPEN SOURCE PROJECTS I CAN HACK ON!

After a bit of a coding break due to social buisiness, I dove right back in to the second part of the file transfer code. Got a passable UI working, and the backend straightend out to the point where it compiles without warnings, but is still braindamaged.

Lots of bugs have been fixed, optimizations made, and the file tranfer UI has been pryed apart from the backend. The new code is going to be good once its done. It has a few fixes and cleanups which should really be put in CVS through a fixup patch right now, but I'm much too lazy for that.

Tommorow I'll probably wake up, work on the horrible UI a bit more, then go right back to getting the backend fully functional. But I'm incredibly tired right now, to the point where writing this diary entry is horrribly challenging, so I'm going to go to sleep now.