I'm happy to report that with the help of RbdPngn, we have good results so far. The previously mentioned segfaults, blamed on Ebits because while EVAS apps weren't failing, Ebit apps were, was incorrect. This turned out to be a problem with the JPEG loader in EVAS, which was interpreting EDB's as JPEGs.After hours of debugging under the supervision of RbdPngn, he made the discovery that 2 critical functions were being called in the wrong order, which he guessed was possibly a compiler optimization problem... and he was right. On this suggestion I upgraded from GCC 3.0.3 to GCC 3.2, rebuilt EVAS and the segfaults disappeared completely. Thanks also goes to tillsan for his input, and raster for his. The only thing left on the segfault list is Elicit, which I'm not too worried about at the moment, and figure could be worked out if I spent the time. EET is still on the non-building list. I looked at a patch Raster found to add fmemopen functionality to FreeBSD, however on inspection this patch simply was an adaptor, allowing fmemopen to simply call FreeBSD's equivilent. So that's out. GNU LibC 2.x is still not a possibility, and no one on the Sun Developer lists is responding to my questions. However, I have found that GNU LibC 1.09.1 _DOES_ have a Solaris port (at least it did in the very early Solaris 2.x days). I'm working now on trying to build it, but it's not gone well so far. Writing my own fmemopen function was another idea, but it's not so easy as I'd hope since fmemopen uses several other functions found only in GLibC, so a nice GNU allowed hi-jack of the code isn't an option. Anyway, there are possibilties at least. Screenshots of DR17 running on Solaris can be found here: http://www.cuddletech.com/dr17-running1.jpg http://www.cuddletech.com/dr17-running2.jpg http://www.cuddletech.com/dr17-running3.jpg
As far as usability goes. I had to patch up FAM to keep it up more than 10 minutes (to fix the infamous /etc/mnttab problem). Once that was fixed, FAM stayed up long enough to hold EFSD in place. However as you can see in the second screenshot above, when trying to add directory handles for .e/desktop/ directories it gets a NULL and craps out. The solution? Wipe out .e/desktop/. The trade off in doing this is that while DR17 runs, the things we all like about DR17 (iconbar, custom layout, custom background) aren't there, so you have to make change things directly in /usr/local/share. Like a good little bug, FAM is at least nice enough to choke and die after the incident, which means not only is EFSD printing a null, it's passing one too. The only other oddity is that when you load a bit in Etcher no images appear from the bit. Odd, because the bits are valid (I can use the in DR17 on Solaris), and Etcher allows me to add images. Just a strange one, but no major problem since I'm not designing bits on Solaris atm anyway. On the plate at the moment is: keep trying to get GlibC 1.09 to build for EET, figure out what this null dir handle crap is about, and look into why Etcher doesn't load image. Also, my new workstation shows up today, so I'll start building on a second machine to refine the procedure and start building patches and pkg's. benr. -- //Ben Rockwood - UNIX Systems Admin //email: [EMAIL PROTECTED] //web: www.cuddletech.com //-> We do what we can, We give what we have, //-> Our doubt is our passion, and our passion is our task, //-> The rest is the madness of Art. //-> -Henry James ------------------------------------------------------- This SF.NET email is sponsored by: SourceForge Enterprise Edition + IBM + LinuxWorld = Something 2 See! http://www.vasoftware.com _______________________________________________ enlightenment-devel mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
