- ntg-pdftex - NTG Mailing Lists

[pdftex-Bugs][4294] issue with isolated/non-isolated page transparency group
by pdftex-bugs＠sarovar.org 17 Nov '09

17 Nov '09

Bugs item #4294, was opened at 2009-09-01 15:43 Status: Open Priority: 3 Submitted By: John Owens (jowens) Assigned to: Nobody (None) Summary: issue with isolated/non-isolated page transparency group Category: PDF inclusion Group: None Resolution: None Initial Comment: (The last bug didn't show up. So I am posting a web link to the three files referenced below: http://tinyurl.com/n4jjm4 ) Howdy, I'm attaching 3 files. One is a tar.gz of a LaTeX document. When we first submitted the paper built from the document, we used dvips and Distiller. Now we're using pdflatex for the final version. Both pdfs are attached. However, strangely, page 4 on the pdflatex pdf, when viewed in Apple Preview on Snow Leopard (10.6), has substantially lighter fonts than the other pages. This is probably Preview's problem, not pdflatex's problem, and I've filed an Apple bug. However, from talking to a PDF expert (summary below), it appears that pdflatex is doing something weird, so I thought I'd bring it to your attention. The base page transparency group for page 4 is nonisolated, where for all the other pages the /I key is not present (and thus they are implicitly isolated). The PDF is correct, although a little weird--there is no point in even specifying the isolation key of a page's transparency group unless it a) contains non-normal blend modes _and_ b) will be rendered in non-device-like contexts (e.g., as an object in a larger piece of artwork). So I would argue that pdflatex should probably not even bother specifying the /I key, especially given that it appears to trigger a bug--or at least a dramatically different code path--in Preview. This _should_ have no effect, as the value of the /I key for the page group is supposed to be ignored (and treated as if it were true). Furthermore even if Preview were ignoring that requirement, the page doesn't have any blend modes, so mathematically it shouldn't be producing a different result. But my guess is this: Preview is rendering page 4 into a nonisolated transparency group using an initial, opaque, all-white backdrop, when it should be completely ignoring the /I key and rendering into a completely transparent backdrop. This shouldn't lead to different results unless a blend mode is present. But,the isolated and nonisolated code paths are somehow sufficiently different that they generate different pixel values--due to precision or perhaps an actual bug. But most of what I say is conjecture. See chapter 7.3.6 ("Page Group") of the PDF Reference Manual, sixth edition. The Distiller version doesn't even have transparency group objects on the pages (which is normal). So I wouldn't expect it to have the same issues in Preview, and it doesn't. (This is LaTeX source for a paper, so though I know you'll need the source to do your investigations and bug fixes, kindly don't do anything else with it beyond that.) $ pdflatex --version pdfTeX 3.1415926-1.40.10-2.2 (TeX Live 2009) kpathsea version 5.0.0 Copyright 2009 Peter Breitenlohner (eTeX)/Han The Thanh (pdfTeX). There is NO warranty. Redistribution of this software is covered by the terms of both the pdfTeX copyright and the Lesser GNU General Public License. For more information about these matters, see the file named COPYING and the pdfTeX source. Primary author of pdfTeX: Peter Breitenlohner (eTeX)/Han The Thanh (pdfTeX). Compiled with libpng 1.2.39; using libpng 1.2.39 Compiled with zlib 1.2.3; using zlib 1.2.3 Compiled with xpdf version 3.02pl3 $ uname -a Darwin classico-2.local 10.0.0 Darwin Kernel Version 10.0.0: Fri Jul 31 22:47:34 PDT 2009; root:xnu-1456.1.25~1/RELEASE_I386 i386 i386 ---------------------------------------------------------------------- Comment By: John Owens (jowens) Date: 2009-11-16 16:33 Message: Quite sensible. Not reasonable to expect pdftex to analyze the PDF and decide if a page's group dictionary is unnecessary. I think you can close this out. Thanks for looking at it! ---------------------------------------------------------------------- Comment By: Martin Schröder (oneiros) Date: 2009-11-15 16:08 Message: John, the display problem is most likely a problem of apple - I can't reproduce the effext with AR9 on Linux. The page group on page 4 comes from the included pdf (figures/multiValue.pdf). pdftex will copy a page group from an included pdf (or synthesize one for pngs with transparency). I don't think it should analyze the content of the page groups it copies. ---------------------------------------------------------------------- Comment By: John Owens (jowens) Date: 2009-11-15 15:46 Message: Just checkin' in. Thanks! ---------------------------------------------------------------------- Comment By: The Thanh Han (hanthethanh) Date: 2009-10-23 09:41 Message: yes this is. Sorry for not replying, I will take a close look in about 2 weeks. ---------------------------------------------------------------------- Comment By: John Owens (jowens) Date: 2009-10-22 13:21 Message: Is this the right place to report pdftex issues? ---------------------------------------------------------------------- You can respond by visiting: http://sarovar.org/tracker/?func=detail&atid=493&aid=4294&group_id=106

1 0

[pdftex-Bugs][4294] issue with isolated/non-isolated page transparency group
by pdftex-bugs＠sarovar.org 16 Nov '09

16 Nov '09

Bugs item #4294, was opened at 2009-09-02 00:43 Status: Open Priority: 3 Submitted By: John Owens (jowens) Assigned to: Nobody (None) Summary: issue with isolated/non-isolated page transparency group >Category: PDF inclusion Group: None Resolution: None Initial Comment: (The last bug didn't show up. So I am posting a web link to the three files referenced below: http://tinyurl.com/n4jjm4 ) Howdy, I'm attaching 3 files. One is a tar.gz of a LaTeX document. When we first submitted the paper built from the document, we used dvips and Distiller. Now we're using pdflatex for the final version. Both pdfs are attached. However, strangely, page 4 on the pdflatex pdf, when viewed in Apple Preview on Snow Leopard (10.6), has substantially lighter fonts than the other pages. This is probably Preview's problem, not pdflatex's problem, and I've filed an Apple bug. However, from talking to a PDF expert (summary below), it appears that pdflatex is doing something weird, so I thought I'd bring it to your attention. The base page transparency group for page 4 is nonisolated, where for all the other pages the /I key is not present (and thus they are implicitly isolated). The PDF is correct, although a little weird--there is no point in even specifying the isolation key of a page's transparency group unless it a) contains non-normal blend modes _and_ b) will be rendered in non-device-like contexts (e.g., as an object in a larger piece of artwork). So I would argue that pdflatex should probably not even bother specifying the /I key, especially given that it appears to trigger a bug--or at least a dramatically different code path--in Preview. This _should_ have no effect, as the value of the /I key for the page group is supposed to be ignored (and treated as if it were true). Furthermore even if Preview were ignoring that requirement, the page doesn't have any blend modes, so mathematically it shouldn't be producing a different result. But my guess is this: Preview is rendering page 4 into a nonisolated transparency group using an initial, opaque, all-white backdrop, when it should be completely ignoring the /I key and rendering into a completely transparent backdrop. This shouldn't lead to different results unless a blend mode is present. But,the isolated and nonisolated code paths are somehow sufficiently different that they generate different pixel values--due to precision or perhaps an actual bug. But most of what I say is conjecture. See chapter 7.3.6 ("Page Group") of the PDF Reference Manual, sixth edition. The Distiller version doesn't even have transparency group objects on the pages (which is normal). So I wouldn't expect it to have the same issues in Preview, and it doesn't. (This is LaTeX source for a paper, so though I know you'll need the source to do your investigations and bug fixes, kindly don't do anything else with it beyond that.) $ pdflatex --version pdfTeX 3.1415926-1.40.10-2.2 (TeX Live 2009) kpathsea version 5.0.0 Copyright 2009 Peter Breitenlohner (eTeX)/Han The Thanh (pdfTeX). There is NO warranty. Redistribution of this software is covered by the terms of both the pdfTeX copyright and the Lesser GNU General Public License. For more information about these matters, see the file named COPYING and the pdfTeX source. Primary author of pdfTeX: Peter Breitenlohner (eTeX)/Han The Thanh (pdfTeX). Compiled with libpng 1.2.39; using libpng 1.2.39 Compiled with zlib 1.2.3; using zlib 1.2.3 Compiled with xpdf version 3.02pl3 $ uname -a Darwin classico-2.local 10.0.0 Darwin Kernel Version 10.0.0: Fri Jul 31 22:47:34 PDT 2009; root:xnu-1456.1.25~1/RELEASE_I386 i386 i386 ---------------------------------------------------------------------- >Comment By: Martin Schröder (oneiros) Date: 2009-11-16 01:08 Message: John, the display problem is most likely a problem of apple - I can't reproduce the effext with AR9 on Linux. The page group on page 4 comes from the included pdf (figures/multiValue.pdf). pdftex will copy a page group from an included pdf (or synthesize one for pngs with transparency). I don't think it should analyze the content of the page groups it copies. ---------------------------------------------------------------------- Comment By: John Owens (jowens) Date: 2009-11-16 00:46 Message: Just checkin' in. Thanks! ---------------------------------------------------------------------- Comment By: The Thanh Han (hanthethanh) Date: 2009-10-23 18:41 Message: yes this is. Sorry for not replying, I will take a close look in about 2 weeks. ---------------------------------------------------------------------- Comment By: John Owens (jowens) Date: 2009-10-22 22:21 Message: Is this the right place to report pdftex issues? ---------------------------------------------------------------------- You can respond by visiting: http://sarovar.org/tracker/?func=detail&atid=493&aid=4294&group_id=106

1 0

[pdftex-Bugs][4294] issue with isolated/non-isolated page transparency group
by pdftex-bugs＠sarovar.org 15 Nov '09

15 Nov '09

Bugs item #4294, was opened at 2009-09-01 15:43 Status: Open Priority: 3 Submitted By: John Owens (jowens) Assigned to: Nobody (None) Summary: issue with isolated/non-isolated page transparency group Category: None Group: None Resolution: None Initial Comment: (The last bug didn't show up. So I am posting a web link to the three files referenced below: http://tinyurl.com/n4jjm4 ) Howdy, I'm attaching 3 files. One is a tar.gz of a LaTeX document. When we first submitted the paper built from the document, we used dvips and Distiller. Now we're using pdflatex for the final version. Both pdfs are attached. However, strangely, page 4 on the pdflatex pdf, when viewed in Apple Preview on Snow Leopard (10.6), has substantially lighter fonts than the other pages. This is probably Preview's problem, not pdflatex's problem, and I've filed an Apple bug. However, from talking to a PDF expert (summary below), it appears that pdflatex is doing something weird, so I thought I'd bring it to your attention. The base page transparency group for page 4 is nonisolated, where for all the other pages the /I key is not present (and thus they are implicitly isolated). The PDF is correct, although a little weird--there is no point in even specifying the isolation key of a page's transparency group unless it a) contains non-normal blend modes _and_ b) will be rendered in non-device-like contexts (e.g., as an object in a larger piece of artwork). So I would argue that pdflatex should probably not even bother specifying the /I key, especially given that it appears to trigger a bug--or at least a dramatically different code path--in Preview. This _should_ have no effect, as the value of the /I key for the page group is supposed to be ignored (and treated as if it were true). Furthermore even if Preview were ignoring that requirement, the page doesn't have any blend modes, so mathematically it shouldn't be producing a different result. But my guess is this: Preview is rendering page 4 into a nonisolated transparency group using an initial, opaque, all-white backdrop, when it should be completely ignoring the /I key and rendering into a completely transparent backdrop. This shouldn't lead to different results unless a blend mode is present. But,the isolated and nonisolated code paths are somehow sufficiently different that they generate different pixel values--due to precision or perhaps an actual bug. But most of what I say is conjecture. See chapter 7.3.6 ("Page Group") of the PDF Reference Manual, sixth edition. The Distiller version doesn't even have transparency group objects on the pages (which is normal). So I wouldn't expect it to have the same issues in Preview, and it doesn't. (This is LaTeX source for a paper, so though I know you'll need the source to do your investigations and bug fixes, kindly don't do anything else with it beyond that.) $ pdflatex --version pdfTeX 3.1415926-1.40.10-2.2 (TeX Live 2009) kpathsea version 5.0.0 Copyright 2009 Peter Breitenlohner (eTeX)/Han The Thanh (pdfTeX). There is NO warranty. Redistribution of this software is covered by the terms of both the pdfTeX copyright and the Lesser GNU General Public License. For more information about these matters, see the file named COPYING and the pdfTeX source. Primary author of pdfTeX: Peter Breitenlohner (eTeX)/Han The Thanh (pdfTeX). Compiled with libpng 1.2.39; using libpng 1.2.39 Compiled with zlib 1.2.3; using zlib 1.2.3 Compiled with xpdf version 3.02pl3 $ uname -a Darwin classico-2.local 10.0.0 Darwin Kernel Version 10.0.0: Fri Jul 31 22:47:34 PDT 2009; root:xnu-1456.1.25~1/RELEASE_I386 i386 i386 ---------------------------------------------------------------------- Comment By: John Owens (jowens) Date: 2009-11-15 15:46 Message: Just checkin' in. Thanks! ---------------------------------------------------------------------- Comment By: The Thanh Han (hanthethanh) Date: 2009-10-23 09:41 Message: yes this is. Sorry for not replying, I will take a close look in about 2 weeks. ---------------------------------------------------------------------- Comment By: John Owens (jowens) Date: 2009-10-22 13:21 Message: Is this the right place to report pdftex issues? ---------------------------------------------------------------------- You can respond by visiting: http://sarovar.org/tracker/?func=detail&atid=493&aid=4294&group_id=106

1 0

[pdftex-Support Requests][4298] pdfTeX mit GCC für Windows kompilieren geht nicht
by pdftex-supportrequests＠sarovar.org 14 Nov '09

14 Nov '09

Support Requests item #4298, was opened at 2009-11-14 20:24 Status: Open Priority: 3 Submitted By: Siggi Siggi (siggi) Assigned to: Nobody (None) Summary: pdfTeX mit GCC für Windows kompilieren geht nicht Category: None Group: None Resolution: None Initial Comment: ch möchte gerne eine schlange Lösung zum automatisierten Erzeugen von PDF-Dateien haben - am besten indem ich .tex-Dateien konvertiere. Da pdfTeX ein schlankes Programm ist, versuchte ich den Quellcode zu kompilieren (da ich keine fertigen Binaries für Windows finden konnte). Ich kompiliere in der Umgebung MSys mit dem GCC. Dabei kommt es zuerst zum Typenkonflikt durch die erneute Deklaration des Datentyp 'Boolean' im Quellcode von pdfTeX. Wenn ich in den GCC includes die typedef auskommentiere, schreitet der Kompilierprozess weiter voran, bricht aber schließlich erneut ab mit der Meldung: http://a.imagehost.org/view/0059/cmd Kann mir jemand sagen, warum ganz unten kein Dateiname vorhanden ist, weshalb der Kompiliervorgang fehlschlägt? Unabhängig davon: Gibt es die Möglichkeit, pdfTeX komplett in eine .exe-Datei oder .dll-Datei zu kompilieren (ohne externe Bibliotheken)? Das wäre eigentlich die optimale Lösung. Grüße siggi ---------------------------------------------------------------------- You can respond by visiting: http://sarovar.org/tracker/?func=detail&atid=494&aid=4298&group_id=106

1 0

[pdftex-Support Requests][4297] conflicting types for 'getline'
by pdftex-supportrequests＠sarovar.org 02 Nov '09

02 Nov '09

Support Requests item #4297, was opened at 2009-10-31 11:50 Status: Open Priority: 3 Submitted By: Bob Tennent (rdtennent) Assigned to: Nobody (None) Summary: conflicting types for 'getline' Category: Install Problem (example) Group: None Resolution: None Initial Comment: Building pdftex on a Fedora 11 system using build.sh fails: gcc -DHAVE_CONFIG_H -I. -I../../../src/texk/web2c -I.. -I../../../src/texk/web2c/.. -I../../libs/obsdcompat -I../../libs/obsdcompat/.. -I../../../src/texk/web2c/../../libs/obsdcompat -I../../../src/texk/web2c/../../libs/obsdcompat/.. -I../../../src/texk/web2c/../../libs/zlib -g -O2 -c tangleboot.c -o tangleboot.o In file included from tangleboot.c:94: tangleboot.h:34: error: conflicting types for 'getline' /usr/include/stdio.h:653: note: previous declaration of 'getline' was here tangleboot.c:2175: error: conflicting types for 'getline' /usr/include/stdio.h:653: note: previous declaration of 'getline' was here I've built pdftex often on other systems without running into this. What's wrong? ---------------------------------------------------------------------- Comment By: Bob Tennent (rdtennent) Date: 2009-11-02 10:05 Message: >| Near the end of texk/web2c/cpascal.h: #ifdef getline #undef getline #endif #define getline web2c_getline Thanks. I'd change this request to "closed" if I could figure out how :+) ---------------------------------------------------------------------- Comment By: Taco Hoekwater (taco) Date: 2009-11-02 09:22 Message: Near the end of texk/web2c/cpascal.h: #ifdef getline #undef getline #endif #define getline web2c_getline ---------------------------------------------------------------------- Comment By: Bob Tennent (rdtennent) Date: 2009-11-02 09:08 Message: "there is a fix for this problem in texlive 2009" That won't help those of us using a distribution like Fedora 11 which is still at texlive-2007. Editing stdio.h isn't an option for most users. And editing the source of pdftex isn't straightforward because of the conversions. What exactly is the "fix" implemented in texlive 2009? ---------------------------------------------------------------------- Comment By: Martin Schröder (oneiros) Date: 2009-11-02 02:30 Message: POSIX 2008? That's new. It seems the glibc team likes to break old programs. http://blog.scottt.tw/2009/04/getline-is-in-posix-2008-and-exposed-in.html and numerous other hits. ---------------------------------------------------------------------- Comment By: Taco Hoekwater (taco) Date: 2009-11-02 02:13 Message: FWIW: there is a fix for this problem in texlive 2009. ---------------------------------------------------------------------- Comment By: Bob Tennent (rdtennent) Date: 2009-11-01 22:03 Message: This is the response at the Fedora bugzilla: getline is a standard POSIX 2008 function, and glibc headers by default offer POSIX 2008 namespace. See info libc 'Feature Test Macros' for details. You should just either fix pdftex to use a different name, not reserved by POSIX 2008, or use a namespace macro that doesn't provide POSIX 2008 stuff. https://bugzilla.redhat.com/show_bug.cgi?id=532230 ---------------------------------------------------------------------- Comment By: Bob Tennent (rdtennent) Date: 2009-10-31 12:50 Message: According to some reports on the web, the problem is in glibc-2.10.1. Apparently the non-standard definition of getline there should be "protected". Indeed, if I comment out that definition in /usr/include/stdio.h, pdftex builds normally. I'm willing to file a bug report in the Fedora bugzilla but I need some help: what exactly is the "protection" needed? ---------------------------------------------------------------------- You can respond by visiting: http://sarovar.org/tracker/?func=detail&atid=494&aid=4297&group_id=106

1 0

[pdftex-Support Requests][4297] conflicting types for 'getline'
by pdftex-supportrequests＠sarovar.org 02 Nov '09

02 Nov '09

Support Requests item #4297, was opened at 2009-10-31 16:50 Status: Open Priority: 3 Submitted By: Bob Tennent (rdtennent) Assigned to: Nobody (None) Summary: conflicting types for 'getline' Category: Install Problem (example) Group: None Resolution: None Initial Comment: Building pdftex on a Fedora 11 system using build.sh fails: gcc -DHAVE_CONFIG_H -I. -I../../../src/texk/web2c -I.. -I../../../src/texk/web2c/.. -I../../libs/obsdcompat -I../../libs/obsdcompat/.. -I../../../src/texk/web2c/../../libs/obsdcompat -I../../../src/texk/web2c/../../libs/obsdcompat/.. -I../../../src/texk/web2c/../../libs/zlib -g -O2 -c tangleboot.c -o tangleboot.o In file included from tangleboot.c:94: tangleboot.h:34: error: conflicting types for 'getline' /usr/include/stdio.h:653: note: previous declaration of 'getline' was here tangleboot.c:2175: error: conflicting types for 'getline' /usr/include/stdio.h:653: note: previous declaration of 'getline' was here I've built pdftex often on other systems without running into this. What's wrong? ---------------------------------------------------------------------- >Comment By: Taco Hoekwater (taco) Date: 2009-11-02 15:22 Message: Near the end of texk/web2c/cpascal.h: #ifdef getline #undef getline #endif #define getline web2c_getline ---------------------------------------------------------------------- Comment By: Bob Tennent (rdtennent) Date: 2009-11-02 15:08 Message: "there is a fix for this problem in texlive 2009" That won't help those of us using a distribution like Fedora 11 which is still at texlive-2007. Editing stdio.h isn't an option for most users. And editing the source of pdftex isn't straightforward because of the conversions. What exactly is the "fix" implemented in texlive 2009? ---------------------------------------------------------------------- Comment By: Martin Schröder (oneiros) Date: 2009-11-02 08:30 Message: POSIX 2008? That's new. It seems the glibc team likes to break old programs. http://blog.scottt.tw/2009/04/getline-is-in-posix-2008-and-exposed-in.html and numerous other hits. ---------------------------------------------------------------------- Comment By: Taco Hoekwater (taco) Date: 2009-11-02 08:13 Message: FWIW: there is a fix for this problem in texlive 2009. ---------------------------------------------------------------------- Comment By: Bob Tennent (rdtennent) Date: 2009-11-02 04:03 Message: This is the response at the Fedora bugzilla: getline is a standard POSIX 2008 function, and glibc headers by default offer POSIX 2008 namespace. See info libc 'Feature Test Macros' for details. You should just either fix pdftex to use a different name, not reserved by POSIX 2008, or use a namespace macro that doesn't provide POSIX 2008 stuff. https://bugzilla.redhat.com/show_bug.cgi?id=532230 ---------------------------------------------------------------------- Comment By: Bob Tennent (rdtennent) Date: 2009-10-31 17:50 Message: According to some reports on the web, the problem is in glibc-2.10.1. Apparently the non-standard definition of getline there should be "protected". Indeed, if I comment out that definition in /usr/include/stdio.h, pdftex builds normally. I'm willing to file a bug report in the Fedora bugzilla but I need some help: what exactly is the "protection" needed? ---------------------------------------------------------------------- You can respond by visiting: http://sarovar.org/tracker/?func=detail&atid=494&aid=4297&group_id=106

1 0

[pdftex-Support Requests][4297] conflicting types for 'getline'
by pdftex-supportrequests＠sarovar.org 02 Nov '09

02 Nov '09

Support Requests item #4297, was opened at 2009-10-31 11:50 Status: Open Priority: 3 Submitted By: Bob Tennent (rdtennent) Assigned to: Nobody (None) Summary: conflicting types for 'getline' Category: Install Problem (example) Group: None Resolution: None Initial Comment: Building pdftex on a Fedora 11 system using build.sh fails: gcc -DHAVE_CONFIG_H -I. -I../../../src/texk/web2c -I.. -I../../../src/texk/web2c/.. -I../../libs/obsdcompat -I../../libs/obsdcompat/.. -I../../../src/texk/web2c/../../libs/obsdcompat -I../../../src/texk/web2c/../../libs/obsdcompat/.. -I../../../src/texk/web2c/../../libs/zlib -g -O2 -c tangleboot.c -o tangleboot.o In file included from tangleboot.c:94: tangleboot.h:34: error: conflicting types for 'getline' /usr/include/stdio.h:653: note: previous declaration of 'getline' was here tangleboot.c:2175: error: conflicting types for 'getline' /usr/include/stdio.h:653: note: previous declaration of 'getline' was here I've built pdftex often on other systems without running into this. What's wrong? ---------------------------------------------------------------------- Comment By: Bob Tennent (rdtennent) Date: 2009-11-02 09:08 Message: "there is a fix for this problem in texlive 2009" That won't help those of us using a distribution like Fedora 11 which is still at texlive-2007. Editing stdio.h isn't an option for most users. And editing the source of pdftex isn't straightforward because of the conversions. What exactly is the "fix" implemented in texlive 2009? ---------------------------------------------------------------------- Comment By: Martin Schröder (oneiros) Date: 2009-11-02 02:30 Message: POSIX 2008? That's new. It seems the glibc team likes to break old programs. http://blog.scottt.tw/2009/04/getline-is-in-posix-2008-and-exposed-in.html and numerous other hits. ---------------------------------------------------------------------- Comment By: Taco Hoekwater (taco) Date: 2009-11-02 02:13 Message: FWIW: there is a fix for this problem in texlive 2009. ---------------------------------------------------------------------- Comment By: Bob Tennent (rdtennent) Date: 2009-11-01 22:03 Message: This is the response at the Fedora bugzilla: getline is a standard POSIX 2008 function, and glibc headers by default offer POSIX 2008 namespace. See info libc 'Feature Test Macros' for details. You should just either fix pdftex to use a different name, not reserved by POSIX 2008, or use a namespace macro that doesn't provide POSIX 2008 stuff. https://bugzilla.redhat.com/show_bug.cgi?id=532230 ---------------------------------------------------------------------- Comment By: Bob Tennent (rdtennent) Date: 2009-10-31 12:50 Message: According to some reports on the web, the problem is in glibc-2.10.1. Apparently the non-standard definition of getline there should be "protected". Indeed, if I comment out that definition in /usr/include/stdio.h, pdftex builds normally. I'm willing to file a bug report in the Fedora bugzilla but I need some help: what exactly is the "protection" needed? ---------------------------------------------------------------------- You can respond by visiting: http://sarovar.org/tracker/?func=detail&atid=494&aid=4297&group_id=106

1 0

[pdftex-Support Requests][4297] conflicting types for 'getline'
by pdftex-supportrequests＠sarovar.org 02 Nov '09

02 Nov '09

Support Requests item #4297, was opened at 2009-10-31 16:50 Status: Open Priority: 3 Submitted By: Bob Tennent (rdtennent) Assigned to: Nobody (None) Summary: conflicting types for 'getline' Category: Install Problem (example) Group: None Resolution: None Initial Comment: Building pdftex on a Fedora 11 system using build.sh fails: gcc -DHAVE_CONFIG_H -I. -I../../../src/texk/web2c -I.. -I../../../src/texk/web2c/.. -I../../libs/obsdcompat -I../../libs/obsdcompat/.. -I../../../src/texk/web2c/../../libs/obsdcompat -I../../../src/texk/web2c/../../libs/obsdcompat/.. -I../../../src/texk/web2c/../../libs/zlib -g -O2 -c tangleboot.c -o tangleboot.o In file included from tangleboot.c:94: tangleboot.h:34: error: conflicting types for 'getline' /usr/include/stdio.h:653: note: previous declaration of 'getline' was here tangleboot.c:2175: error: conflicting types for 'getline' /usr/include/stdio.h:653: note: previous declaration of 'getline' was here I've built pdftex often on other systems without running into this. What's wrong? ---------------------------------------------------------------------- >Comment By: Martin Schröder (oneiros) Date: 2009-11-02 08:30 Message: POSIX 2008? That's new. It seems the glibc team likes to break old programs. http://blog.scottt.tw/2009/04/getline-is-in-posix-2008-and-exposed-in.html and numerous other hits. ---------------------------------------------------------------------- Comment By: Taco Hoekwater (taco) Date: 2009-11-02 08:13 Message: FWIW: there is a fix for this problem in texlive 2009. ---------------------------------------------------------------------- Comment By: Bob Tennent (rdtennent) Date: 2009-11-02 04:03 Message: This is the response at the Fedora bugzilla: getline is a standard POSIX 2008 function, and glibc headers by default offer POSIX 2008 namespace. See info libc 'Feature Test Macros' for details. You should just either fix pdftex to use a different name, not reserved by POSIX 2008, or use a namespace macro that doesn't provide POSIX 2008 stuff. https://bugzilla.redhat.com/show_bug.cgi?id=532230 ---------------------------------------------------------------------- Comment By: Bob Tennent (rdtennent) Date: 2009-10-31 17:50 Message: According to some reports on the web, the problem is in glibc-2.10.1. Apparently the non-standard definition of getline there should be "protected". Indeed, if I comment out that definition in /usr/include/stdio.h, pdftex builds normally. I'm willing to file a bug report in the Fedora bugzilla but I need some help: what exactly is the "protection" needed? ---------------------------------------------------------------------- You can respond by visiting: http://sarovar.org/tracker/?func=detail&atid=494&aid=4297&group_id=106

1 0

[pdftex-Support Requests][4297] conflicting types for 'getline'
by pdftex-supportrequests＠sarovar.org 02 Nov '09

02 Nov '09

Support Requests item #4297, was opened at 2009-10-31 16:50 Status: Open Priority: 3 Submitted By: Bob Tennent (rdtennent) Assigned to: Nobody (None) Summary: conflicting types for 'getline' Category: Install Problem (example) Group: None Resolution: None Initial Comment: Building pdftex on a Fedora 11 system using build.sh fails: gcc -DHAVE_CONFIG_H -I. -I../../../src/texk/web2c -I.. -I../../../src/texk/web2c/.. -I../../libs/obsdcompat -I../../libs/obsdcompat/.. -I../../../src/texk/web2c/../../libs/obsdcompat -I../../../src/texk/web2c/../../libs/obsdcompat/.. -I../../../src/texk/web2c/../../libs/zlib -g -O2 -c tangleboot.c -o tangleboot.o In file included from tangleboot.c:94: tangleboot.h:34: error: conflicting types for 'getline' /usr/include/stdio.h:653: note: previous declaration of 'getline' was here tangleboot.c:2175: error: conflicting types for 'getline' /usr/include/stdio.h:653: note: previous declaration of 'getline' was here I've built pdftex often on other systems without running into this. What's wrong? ---------------------------------------------------------------------- >Comment By: Taco Hoekwater (taco) Date: 2009-11-02 08:13 Message: FWIW: there is a fix for this problem in texlive 2009. ---------------------------------------------------------------------- Comment By: Bob Tennent (rdtennent) Date: 2009-11-02 04:03 Message: This is the response at the Fedora bugzilla: getline is a standard POSIX 2008 function, and glibc headers by default offer POSIX 2008 namespace. See info libc 'Feature Test Macros' for details. You should just either fix pdftex to use a different name, not reserved by POSIX 2008, or use a namespace macro that doesn't provide POSIX 2008 stuff. https://bugzilla.redhat.com/show_bug.cgi?id=532230 ---------------------------------------------------------------------- Comment By: Bob Tennent (rdtennent) Date: 2009-10-31 17:50 Message: According to some reports on the web, the problem is in glibc-2.10.1. Apparently the non-standard definition of getline there should be "protected". Indeed, if I comment out that definition in /usr/include/stdio.h, pdftex builds normally. I'm willing to file a bug report in the Fedora bugzilla but I need some help: what exactly is the "protection" needed? ---------------------------------------------------------------------- You can respond by visiting: http://sarovar.org/tracker/?func=detail&atid=494&aid=4297&group_id=106

1 0

[pdftex-Support Requests][4297] conflicting types for 'getline'
by pdftex-supportrequests＠sarovar.org 02 Nov '09

02 Nov '09

Support Requests item #4297, was opened at 2009-10-31 11:50 Status: Open Priority: 3 Submitted By: Bob Tennent (rdtennent) Assigned to: Nobody (None) Summary: conflicting types for 'getline' Category: Install Problem (example) Group: None Resolution: None Initial Comment: Building pdftex on a Fedora 11 system using build.sh fails: gcc -DHAVE_CONFIG_H -I. -I../../../src/texk/web2c -I.. -I../../../src/texk/web2c/.. -I../../libs/obsdcompat -I../../libs/obsdcompat/.. -I../../../src/texk/web2c/../../libs/obsdcompat -I../../../src/texk/web2c/../../libs/obsdcompat/.. -I../../../src/texk/web2c/../../libs/zlib -g -O2 -c tangleboot.c -o tangleboot.o In file included from tangleboot.c:94: tangleboot.h:34: error: conflicting types for 'getline' /usr/include/stdio.h:653: note: previous declaration of 'getline' was here tangleboot.c:2175: error: conflicting types for 'getline' /usr/include/stdio.h:653: note: previous declaration of 'getline' was here I've built pdftex often on other systems without running into this. What's wrong? ---------------------------------------------------------------------- Comment By: Bob Tennent (rdtennent) Date: 2009-11-01 22:03 Message: This is the response at the Fedora bugzilla: getline is a standard POSIX 2008 function, and glibc headers by default offer POSIX 2008 namespace. See info libc 'Feature Test Macros' for details. You should just either fix pdftex to use a different name, not reserved by POSIX 2008, or use a namespace macro that doesn't provide POSIX 2008 stuff. https://bugzilla.redhat.com/show_bug.cgi?id=532230 ---------------------------------------------------------------------- Comment By: Bob Tennent (rdtennent) Date: 2009-10-31 12:50 Message: According to some reports on the web, the problem is in glibc-2.10.1. Apparently the non-standard definition of getline there should be "protected". Indeed, if I comment out that definition in /usr/include/stdio.h, pdftex builds normally. I'm willing to file a bug report in the Fedora bugzilla but I need some help: what exactly is the "protection" needed? ---------------------------------------------------------------------- You can respond by visiting: http://sarovar.org/tracker/?func=detail&atid=494&aid=4297&group_id=106

1 0