Re: 2.99.12 (Irix failure)
my manpage (IRIX-6.5.24) says, you have to include libgen.h You are correct... But you also need to link with -lgen... Therefore, to use it - you would need a test in texk/web2c/configure - and then link in pdftexdir with it as well as ttf2afm in texk/web2c directory So - either use xbasename - or redo configure and two Makefile.in a bit... Either would work for me... But note that ttf2afm.c is already using xstrdup right over there as well instead of testing for it in the system libraries - so the precedence would be to use xbasename Ezra
Re: 2.99.11: Compilation under irix broken
Well... New problem now... TEXINPUTS=.::$TEXINPUTS \ /usr/sbin/perl ../texi2html -V ./texi2html.texi Can't locate Data/Dumper.pm in @INC (@INC contains: /usr/share/lib/perl5/irix-n32/5.00405 /usr/share/lib/perl5 /usr/share/lib/perl5/site_perl/irix-n32 /usr/share/lib/perl5/site_perl /usr/share/lib/perl5/sgi_perl .) at ../texi2html line 6142. If I comment out the use Data::Dumper, I get... (which when tested under linux shows to be okay) TEXINPUTS=.::$TEXINPUTS \ /usr/sbin/perl ../texi2html -V ./texi2html.texi exists operator argument is not a HASH element at ../texi2html line 16364. Ezra
Re: texi2html new try
sungritzmann4 % ./texi2html ../doc/texi2html.texi exists operator argument is not a HASH element at ./texi2html line 16364. I have tested the new new version and it works now... Yay... And the resulting html file looks good in a browser (this is w/ perl 5.00405) Hmm - tested on an older sgi running 6.5.16m - which is perl version 5.00404 (vs. 5.00405) - that one is missing File::Spec... Perhaps configure could test for File::Spec? I guess how far back do you really want this to work... Looks like you cannot download that version of perl anymore... Either that, or require 5.00405 to function... Maybe test for that version in the configure... Ezra
Re: 2.99.12 (Irix failure)
Have a problem under Irix... basename is not present on the system Need to use xbasename.. (And one needs to remove texk/web2c/ttf2afm.c before recompiling as it is copied) I'll let you know if there any other problems by tomorrow morning... *** texk/web2c/pdftexdir/ttf2afm.c 2005/02/04 03:24:02 1.1 --- texk/web2c/pdftexdir/ttf2afm.c 2005/02/04 03:24:31 *** *** 975,981 cur_file_name = argv[optind]; if (print_cmap) { buf = xstrdup(cur_file_name); ! bname = basename(buf); if ((s = rindex(bname, '.')) != NULL) *s = 0; } --- 975,981 cur_file_name = argv[optind]; if (print_cmap) { buf = xstrdup(cur_file_name); ! bname = xbasename(buf); if ((s = rindex(bname, '.')) != NULL) *s = 0; }
Re: 2.99.7: first rc of teTeX-3.0
Initial tests show that font creation is going into the share/texmf tree now. That is intentional, because TEXMFVAR points there. Font creation used to go to the tree of the font source (if that was writable), but to fit into the general concept, I have changed that (it can be undone by removing texmfvar from the file mktex.cnf). Is there any problem with this approach? Nope - that is what I wanted to see... Earlier pre-releases they went into texmf-dist - which did not make as much sense. Ezra
Re: 2.99.5 release
Compiles aok under Irix 6.5.20m - using native compilers and GNUmake. (system make bombs out in texk somewhere)... Simple tests: xdvi, latex, mpost, pdflatex appears to work. Some comments: Suggestion - in the INSTALL file - it says to type ls | wc -l - I get 27. Perhaps ls -1 | wc -l would be more appropriate to confirm the number of binaries. I notice that when I am setting paper types in texconfig - it is realtively fast (xdvi, dvips) until I set the paper type in PDFTEX. This forces a rebuild of the formats... Looks like most of them (this is even though I leave it set to the default pattern). It is not clear to me why setting the paper size for pdftex rebuilds all the formats (including latex). Perhaps the menu item should indicate tex configuration instead of pdftex. (I notice that latex is a symlink to pdfetex - so it makes sense that all formates are rebuilt). Ezra
Re: [2.99.3] ./configure[597]: syntax error at line 74 : `{' unmatched
I found that the system /bin/sh is sufficient to configure under Irix... I would suggest trying that instead of explicitly using bash... (How did you invoke configure - to change shells?) Ezra
2.99.3 texconfig font dir selection
I have installed 2.99.3 with the default configuration... Running texconfig, I select fonts. In fonts - the top of the menu indicates that files will be stored in share/texmf-dist/... I set vardir to /var/tmp/texfonts Top still indicates texmf-dist will be used. Set varfonts option on -- top of dialog screen says /var/tmp/texfonts/... as expected. Select FONTRO Font permissions are changed in texmf/ --- not texmf-dist... I turn off varfonts option, exit texconfig (after making sure global rw allowed), and run allcm... Fonts are placed in texmf-dist - not texmf... So - it looks like global RO or global RW is going to the wrong directory... (Unless the default configuration is wrong and texmf/fonts should be the location of the fonts - which makes sense to me)...
Re: 2.99.2 release
Found a problem compiling under Irix 6.5.20 using native compiler... libs/gd/gd.c has the following type of errors: cc-1077 cc: ERROR File = gd.c, Line = 2719 The indicated declaration has no storage class or type specifier. inline static void gdImageSetAAPixelColor(gdImagePtr im, int x, int y, int color, int t); ^ This is fixed in newer versions of the gd library... But here is the fix... (essentially remove inline). I will let you know of other problems as discovered... Ezra *** gd.c2004/11/07 16:23:13 1.1 --- gd.c2004/11/07 16:24:53 *** *** 2716,2722 } } ! inline static void gdImageSetAAPixelColor(gdImagePtr im, int x, int y, int color, int t); BGD_DECLARE(void) gdImageSetStyle (gdImagePtr im, int *style, int noOfPixels) { --- 2716,2722 } } ! static void gdImageSetAAPixelColor(gdImagePtr im, int x, int y, int color, int t); BGD_DECLARE(void) gdImageSetStyle (gdImagePtr im, int *style, int noOfPixels) { *** *** 2969,2975 #define BLEND_COLOR(a, nc, c, cc) \ nc = (cc) + (c) - (cc)) * (a)) + c) - (cc)) * (a)) 8) + 0x80) 8); ! inline static void gdImageSetAAPixelColor(gdImagePtr im, int x, int y, int color, int t) { int dr,dg,db,p,r,g,b; p = gdImageGetPixel(im,x,y); --- 2969,2975 #define BLEND_COLOR(a, nc, c, cc) \ nc = (cc) + (c) - (cc)) * (a)) + c) - (cc)) * (a)) 8) + 0x80) 8); ! static void gdImageSetAAPixelColor(gdImagePtr im, int x, int y, int color, int t) { int dr,dg,db,p,r,g,b; p = gdImageGetPixel(im,x,y);
.99.2 release Irix problems with dvipng
There are several issues with the use using a void * in arithmetic. Essentially - the code assumes that a (void * ) + 1 is the same as a (char *) + 1. I am not sure that is totally legal as when adding to a pointer - when moves by the size of what one is pointing to. The Irix error is: The expression must be a pointer to a complete object type. The following patches apply to the dvipng directory... *** color.c 2004/11/07 23:36:50 1.1 --- color.c 2004/11/07 23:38:08 *** *** 66,72 if ((colornames = malloc(fmmap.size*2))==NULL) Fatal(cannot alloc space for color names); pos=fmmap.mmap; ! max=fmmap.mmap+fmmap.size; tmp=colornames; buf=(char*)colornames+fmmap.size; while (posmax *pos!='\\') pos++; --- 66,72 if ((colornames = malloc(fmmap.size*2))==NULL) Fatal(cannot alloc space for color names); pos=fmmap.mmap; ! max=(char *)fmmap.mmap+fmmap.size; tmp=colornames; buf=(char*)colornames+fmmap.size; while (posmax *pos!='\\') pos++; *** enc.c 2004/11/07 23:43:02 1.1 --- enc.c 2004/11/07 23:44:27 *** *** 57,63 encp-name=(char*)encp+sizeof(struct encoding); strcpy(encp-name,encoding); pos=fmmap.mmap; ! max=fmmap.mmap+fmmap.size; buf=encp-name+strlen(encoding)+1; #define SKIPCOMMENT(x) if (*x=='%') while (xmax *x!='\n') x++; while(posmax *pos!='/') { --- 57,63 encp-name=(char*)encp+sizeof(struct encoding); strcpy(encp-name,encoding); pos=fmmap.mmap; ! max=(char *)fmmap.mmap+fmmap.size; buf=encp-name+strlen(encoding)+1; #define SKIPCOMMENT(x) if (*x=='%') while (xmax *x!='\n') x++; while(posmax *pos!='/') { *** fontmap.c 2004/11/07 23:46:05 1.1 --- fontmap.c 2004/11/07 23:46:19 *** *** 73,79 return; } pos = psfont_mmap.mmap; ! end = psfont_mmap.mmap+psfont_mmap.size; while(posend) { while(pos end (*pos=='\n' || *pos==' ' || *pos=='\t' || *pos=='%' || *pos=='*' || *pos==';' || *pos=='#')) { --- 73,79 return; } pos = psfont_mmap.mmap; ! end = (char *)psfont_mmap.mmap+psfont_mmap.size; while(posend) { while(pos end (*pos=='\n' || *pos==' ' || *pos=='\t' || *pos=='%' || *pos=='*' || *pos==';' || *pos=='#')) { *** pk.c2004/11/07 23:39:24 1.1 --- pk.c2004/11/07 23:41:25 *** *** 201,207 ptr-w = shrunk_width; ptr-h = shrunk_height; pos+=n; ! buffer = (char*)alloca(shrunk_width*shrunk_height* shrinkfactor*shrinkfactor*sizeof(char)); (void)memset(buffer,0,shrunk_width*shrunk_height* shrinkfactor*shrinkfactor*sizeof(char)); --- 201,207 ptr-w = shrunk_width; ptr-h = shrunk_height; pos+=n; ! buffer = (unsigned char*)alloca(shrunk_width*shrunk_height* shrinkfactor*shrinkfactor*sizeof(char)); (void)memset(buffer,0,shrunk_width*shrunk_height* shrinkfactor*shrinkfactor*sizeof(char)); *** *** 353,359 position = skip_specials(position); while (*position != PK_POST) { DEBUG_PRINT(DEBUG_PK,(\n @%ld PK CHAR:\t%d, ! (long)((void*)position - tfontp-fmmap.mmap), *position)); if ((tcharptr = malloc(sizeof(struct char_entry))) == NULL) Fatal(can't malloc space for char_entry); tcharptr-flag_byte = *position; --- 353,360 position = skip_specials(position); while (*position != PK_POST) { DEBUG_PRINT(DEBUG_PK,(\n @%ld PK CHAR:\t%d, ! (long)(position - (unsigned char *)tfontp-fmmap.mmap), !*position)); if ((tcharptr = malloc(sizeof(struct char_entry))) == NULL) Fatal(can't malloc space for char_entry); tcharptr-flag_byte = *position; *** tfm.c 2004/11/07 23:47:36 1.1 --- tfm.c 2004/11/07 23:49:01 *** *** 39,52 DEBUG_PRINT((DEBUG_DVI|DEBUG_FT|DEBUG_TFM), (\n OPEN METRICS:\t'%s', tfmname)); if (MmapFile(tfmname,fmmap)) return(false); ! lh = UNumRead(fmmap.mmap+2,2); ! bc = UNumRead(fmmap.mmap+4,2); ! ec = UNumRead(fmmap.mmap+6,2); ! nw = UNumRead(fmmap.mmap+8,2); DEBUG_PRINT(DEBUG_TFM,( %d %d %d %d,lh,bc,ec,nw)); width=alloca(nw*sizeof(dviunits)); c=0; ! position=fmmap.mmap+24+(lh+ec-bc+1)*4; while( c nw ) { width[c] = SNumRead(position,4); c++; --- 39,52 DEBUG_PRINT((DEBUG_DVI|DEBUG_FT|DEBUG_TFM), (\n OPEN METRICS:\t'%s', tfmname)); if (MmapFile(tfmname,fmmap)) return(false); ! lh = UNumRead((unsigned char *)fmmap.mmap+2,2); ! bc = UNumRead((unsigned char *)fmmap.mmap+4,2); ! ec = UNumRead((unsigned char *)fmmap.mmap+6,2); ! nw = UNumRead((unsigned char *)fmmap.mmap+8,2); DEBUG_PRINT(DEBUG_TFM,( %d %d %d %d,lh,bc,ec,nw)); width=alloca(nw*sizeof(dviunits)); c=0; ! position=(unsigned
tetex 2.96 permission problem....
The permissions on utils/texinfo/install-sh is not executable... Not a problem unless your system install is a SysV based one (Irix) and you need to invoke the shell script in utils/texinfo...