Hello, Few days ago there were new windows binaries on Sourceforge - version 0.98.5.3. Yesterday I also found them in Google cache (see the attachment - I am interested mainly in Python 2.6). But they are not now on Sourceforge. Is there a good reason for it (for example there could be problematic)? Or is it a Sourceforge mistake (I am not sure but I think that design of the Sourceforge download page was different)? Maybe there is another explanation - but if it is possible, could windows binaries be uploaded again? Thank you, Petr Marhoun
On Fri, May 29, 2009 at 9:44 PM, Petr Marhoun <pet...@gm...> wrote: > Hello, > > Few days ago there were new windows binaries on Sourceforge - version > 0.98.5.3. Yesterday I also found them in Google cache (see the > attachment - I am interested mainly in Python 2.6). > > But they are not now on Sourceforge. Is there a good reason for it > (for example there could be problematic)? Or is it a Sourceforge > mistake (I am not sure but I think that design of the Sourceforge > download page was different)? > > Maybe there is another explanation - but if it is possible, could > windows binaries be uploaded again? The 0.98.5.3 binaries have a problem with the PNG output, so I pulled them. Charile, do you still have the 0.98.5.2 win32 binaries to reupload while we are sorting out this problem? JDH
I can't say that I have them. Petr, you can use 98.5.3 binaries found here, "http://drop.io/tvuqe3o". Just keep in mind that png operations will fail. On Sat, May 30, 2009 at 2:14 PM, John Hunter <jd...@gm...> wrote: > On Fri, May 29, 2009 at 9:44 PM, Petr Marhoun <pet...@gm...> wrote: >> Hello, >> >> Few days ago there were new windows binaries on Sourceforge - version >> 0.98.5.3. Yesterday I also found them in Google cache (see the >> attachment - I am interested mainly in Python 2.6). >> >> But they are not now on Sourceforge. Is there a good reason for it >> (for example there could be problematic)? Or is it a Sourceforge >> mistake (I am not sure but I think that design of the Sourceforge >> download page was different)? >> >> Maybe there is another explanation - but if it is possible, could >> windows binaries be uploaded again? > > The 0.98.5.3 binaries have a problem with the PNG output, so I pulled > them. Charile, do you still have the 0.98.5.2 win32 binaries to > reupload while we are sorting out this problem? > > JDH >
On Tue, Jun 2, 2009 at 7:59 AM, Charlie Moad <cw...@gm...> wrote: > I can't say that I have them. Petr, you can use 98.5.3 binaries found > here, "http://drop.io/tvuqe3o". Just keep in mind that png operations > will fail. I've done some digging on this -- the segfault is happening in _png.cpp on the line png_write_info(png_ptr, info_ptr); and the error is Unhandled exception at 0x7c91b1fa in python.exe: 0xC0000005: Access violation writing location 0x00000010. when I paste this part of the message into google "python.exe: 0xC0000005: Access violation writing location 0x00000010" I find all kinds of matches. Eg, this page http://stackoverflow.com/questions/391917/jpeg-support-with-ijg-getting-access-violation suggests that there could be a problem in trying to pass FILE * pointers to shared libraries. It seems we've done what is necessary to compile these statically, but I'm just throwing this out there in case it stimulates some ideas. I'm stuck so far in trying to fix this. JDH