Re: Still can't build 21.0-b46 on Solaris

On Sun, Jul 05, 1998 at 11:01:34PM -0700, SL Baur wrote: Sounds like a little/big endianness fuckup for the pixel values when building the local XImage.... ...yup, it's that. glyphs.c conditionalizes the byte order of pixel bytes in the image on WORDS_BIGENDIAN for TrueColor visuals. The problem is that it must follow the _server_ endianness, not the client's one. Untested (I don't have the required configuration) patch follows. OG. 1998-07-06 Olivier Galibert <galibert(a)pobox.com&gt; * glyphs-x.c (convert_EImage_to_XImage): Fix pixel writing problem when the X server endianness is different than the client's one. Index: src/glyphs-x.c =================================================================== RCS file: /usr/CVSroot/XEmacs/xemacs-20/src/glyphs-x.c,v retrieving revision 1.48 diff -u -r1.48 glyphs-x.c --- glyphs-x.c 1998/06/13 04:28:50 1.48 +++ glyphs-x.c 1998/07/06 15:15:06 ＠＠ -198,11 +198,10 ＠＠ gr = *ip++; bl = *ip++; conv.val = pixarray[QUANT_GET_COLOR(qtable,rd,gr,bl)]; -#ifdef WORDS_BIGENDIAN - for (q = 4-byte_cnt; q < 4; q++) *dp++ = conv.cp[q]; -#else - for (q = 0; q < byte_cnt; q++) *dp++ = conv.cp[q]; -#endif + if (outimg->byte_order == MSBFirst) + for (q = 4-byte_cnt; q < 4; q++) *dp++ = conv.cp[q]; + else + for (q = 0; q < byte_cnt; q++) *dp++ = conv.cp[q]; } } xfree(qtable); ＠＠ -267,11 +266,10 ＠＠ bl = *ip++ >> (8 - bbits); conv.val = (rd << rshift) | (gr << gshift) | (bl << bshift); -#ifdef WORDS_BIGENDIAN - for (q = 4-byte_cnt; q < 4; q++) *dp++ = conv.cp[q]; -#else - for (q = 0; q < byte_cnt; q++) *dp++ = conv.cp[q]; -#endif + if (outimg->byte_order == MSBFirst) + for (q = 4-byte_cnt; q < 4; q++) *dp++ = conv.cp[q]; + else + for (q = 0; q < byte_cnt; q++) *dp++ = conv.cp[q]; } } }