Re: Bug somewhere between cygwin, gcc and cvs
Hello All! The newest cygwin DLL version (from 12/25) helped. I've recompiled with -O2, and now it works fine. Thank you very, very much. You may be missing the fact that saying "a problem with opening a directory" is not equivalent to saying "Cygwin isn't supposed to work this way." Pierre Humblet submitted a patch to fix a problem opening directories on Windows 9x/Me last week. So, as always, trying a new cygwin snapshot is a good idea. cgf -- Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple Bug reporting: http://cygwin.com/bugs.html Documentation: http://cygwin.com/docs.html FAQ: http://cygwin.com/faq/
Re: Bug somewhere between cygwin, gcc and cvs
On Wed, Dec 25, 2002 at 03:06:33PM -0500, Igor Pechtchanski wrote: >On Wed, 25 Dec 2002, Christopher Faylor wrote: > >> On Wed, Dec 25, 2002 at 02:29:17PM -0500, Arkadiy Belousov wrote: >> >The version currently available does not support specifying port in >> >CVSROOT. And I need that for ssh tunneling. >> > >> >I do realize I am on the bleeding edge here. Still, it's something to >> >look at when CVS 1.11.2 is ported. Personally, I think it's the >> >compiler. Let's wait till gcc loses (prerelease) label on it. >> >> Despite appearances to the contrary, cygwin's gcc is based on the >> released version of gcc 3.2. There is no better version being planned. >> >> It sounds to me like this is a problem with opening a directory using >> open, as was previously suggested. > >If I understand the documentation at >http://www.opengroup.org/onlinepubs/007904975/functions/open.html >corectly, open()ing a directory is not explicitly prohibited (in fact, the >phrasing of one error value, > >[EISDIR] >The named file is a directory and oflag includes O_WRONLY or O_RDWR. > >suggests that it is allowed). Besides, the code only uses the file >descriptor in a subsequent call to fchdir()... Am I missing something >here? You may be missing the fact that saying "a problem with opening a directory" is not equivalent to saying "Cygwin isn't supposed to work this way." Pierre Humblet submitted a patch to fix a problem opening directories on Windows 9x/Me last week. So, as always, trying a new cygwin snapshot is a good idea. cgf -- Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple Bug reporting: http://cygwin.com/bugs.html Documentation: http://cygwin.com/docs.html FAQ: http://cygwin.com/faq/
Re: Bug somewhere between cygwin, gcc and cvs
On Wed, 25 Dec 2002, Christopher Faylor wrote: > On Wed, Dec 25, 2002 at 02:29:17PM -0500, Arkadiy Belousov wrote: > >The version currently available does not support specifying port in > >CVSROOT. And I need that for ssh tunneling. > > > >I do realize I am on the bleeding edge here. Still, it's something to > >look at when CVS 1.11.2 is ported. Personally, I think it's the > >compiler. Let's wait till gcc loses (prerelease) label on it. > > Despite appearances to the contrary, cygwin's gcc is based on the > released version of gcc 3.2. There is no better version being planned. > > It sounds to me like this is a problem with opening a directory using > open, as was previously suggested. > > cgf If I understand the documentation at http://www.opengroup.org/onlinepubs/007904975/functions/open.html corectly, open()ing a directory is not explicitly prohibited (in fact, the phrasing of one error value, [EISDIR] The named file is a directory and oflag includes O_WRONLY or O_RDWR. suggests that it is allowed). Besides, the code only uses the file descriptor in a subsequent call to fchdir()... Am I missing something here? Igor -- http://cs.nyu.edu/~pechtcha/ |\ _,,,---,,_[EMAIL PROTECTED] ZZZzz /,`.-'`'-. ;-;;,_[EMAIL PROTECTED] |,4- ) )-,_. ,\ ( `'-' Igor Pechtchanski '---''(_/--' `-'\_) fL a.k.a JaguaR-R-R-r-r-r-.-.-. Meow! Oh, boy, virtual memory! Now I'm gonna make myself a really *big* RAMdisk! -- /usr/games/fortune -- Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple Bug reporting: http://cygwin.com/bugs.html Documentation: http://cygwin.com/docs.html FAQ: http://cygwin.com/faq/
Re: Bug somewhere between cygwin, gcc and cvs
On Wed, Dec 25, 2002 at 02:29:17PM -0500, Arkadiy Belousov wrote: >The version currently available does not support specifying port in >CVSROOT. And I need that for ssh tunneling. > >I do realize I am on the bleeding edge here. Still, it's something to >look at when CVS 1.11.2 is ported. Personally, I think it's the >compiler. Let's wait till gcc loses (prerelease) label on it. Despite appearances to the contrary, cygwin's gcc is based on the released version of gcc 3.2. There is no better version being planned. It sounds to me like this is a problem with opening a directory using open, as was previously suggested. cgf -- Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple Bug reporting: http://cygwin.com/bugs.html Documentation: http://cygwin.com/docs.html FAQ: http://cygwin.com/faq/
Re: Bug somewhere between cygwin, gcc and cvs
The version currently available does not support specifying port in CVSROOT. And I need that for ssh tunneling. I do realize I am on the bleeding edge here. Still, it's something to look at when CVS 1.11.2 is ported. Personally, I think it's the compiler. Let's wait till gcc loses (prerelease) label on it. Michael A Chase wrote: On Wed, 25 Dec 2002 13:22:17 -0500 Arkadiy Belousov <[EMAIL PROTECTED]> wrote: I downloaded latest stable CVS code (1.11.2) and compiled it under cygwin. "cvs co", "cvs login" and "cvs commit" worked, but not "cvs update". Since you are trying to build a version of CVS in advance of the version available on Cygwin mirrors, you may not get much help here. I suggest you download the source for the Cygwin CVS (release/cvs/cvs-1.11.0-1-src.tar.gz) and see what is done in the corresponding places there. If that version has the same problems, I'm sure the Cygwin maintainer would appreciate a patch. -- Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple Bug reporting: http://cygwin.com/bugs.html Documentation: http://cygwin.com/docs.html FAQ: http://cygwin.com/faq/
Re: Bug somewhere between cygwin, gcc and cvs
Arkadiy, My information may be out-of-date, but I thought direct access to directories was not POSIX-compliant? Of course, if that's true, one would think CVS would long ago have adapted to that restriction. Anyway, trying to access a directory using any old naive program (i.e., one that doesn't detect and make a special accommodation for directories) such as "dd," "od," or "cat" gives "Is a directory" diagnostics every time. Are you sure your CVS build environment was properly configured? Is the main-line CVS source code featured for (ported to) Cygwin? You said you're trying to build the "latest stable CVS code" and that it's version 1.11.2. However, by my observation, the latest Cygwin CVS is 1.11.0. I'm guessing that the CVS source code you're using does not bear the requisite Cygwin-specific patches. Why don't you use the source code that's distributed with / by the Cygwin project? Of course, unless you're going to modify it, you'll just end up with the functional equivalent of the current Cygwin binary package for CVS. If there's a feature of CVS 1.11.2 that you need, you could probably get a considerable head start by taking the Cygwin patches from the Cygwin 1.11.0 source package and applying them (with any necessary changes, of course) to the 1.11.2 source code. I hope the Cygwin CVS package maintainer will definitively confirm, refute or refine the speculation I've indulged in here... Randall Schulz At 10:22 2002-12-25, Arkadiy Belousov wrote: Hello all! I downloaded latest stable CVS code (1.11.2) and compiled it under cygwin. "cvs co", "cvs login" and "cvs commit" worked, but not "cvs update". I traced the bug to call to open (".", O_RDONLY); at lib/savecwd.c:61 in CVS source. This call is made the same way during commit and update, yet it fails during update only, with the error "file already exist". When I tried to make similar call from other places in the code, I learned that placing the call right before save_cwd is called from recurse.c makes the entire thing work. The next thing I tried was optimization level in the compiler. When I replaced -O2 with -O, the whole situation went away. I am running Windows ME. GCC gives the following in response to -v : $ gcc -v Reading specs from /usr/lib/gcc-lib/i686-pc-cygwin/3.2/specs Configured with: /netrel/src/gcc-3.2-3/configure --enable-languages=c,c++,f77,java --enable-libgcj --enable-threads=posix --with-system-zlib --enable-nls --without-included-gettext --enable-interpreter --disable-sjlj-exceptions --disable-version-specific-runtime-libs --enable-shared --build=i686-pc-linux --host=i686-pc-cygwin --target=i686-pc-cygwin --enable-haifa --prefix=/usr --exec-prefix=/usr --sysconfdir=/etc --libdir=/usr/lib --includedir=/nonexistent/include --libexecd ir=/usr/sbin Thread model: posix gcc version 3.2 20020927 (prerelease) cygwin version is cygwin 1.3.17-1 , according to cygcheck -- Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple Bug reporting: http://cygwin.com/bugs.html Documentation: http://cygwin.com/docs.html FAQ: http://cygwin.com/faq/
Re: Bug somewhere between cygwin, gcc and cvs
On Wed, 25 Dec 2002 13:22:17 -0500 Arkadiy Belousov <[EMAIL PROTECTED]> wrote: > I downloaded latest stable CVS code (1.11.2) and compiled it under cygwin. > "cvs co", "cvs login" and "cvs commit" worked, but not "cvs update". Since you are trying to build a version of CVS in advance of the version available on Cygwin mirrors, you may not get much help here. I suggest you download the source for the Cygwin CVS (release/cvs/cvs-1.11.0-1-src.tar.gz) and see what is done in the corresponding places there. If that version has the same problems, I'm sure the Cygwin maintainer would appreciate a patch. -- Mac :}) ** I normally forward private questions to the appropriate mail list. ** Ask Smarter: http://www.tuxedo.org/~esr/faqs/smart-questions.html Give a hobbit a fish and he eats fish for a day. Give a hobbit a ring and he eats fish for an age. -- Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple Bug reporting: http://cygwin.com/bugs.html Documentation: http://cygwin.com/docs.html FAQ: http://cygwin.com/faq/
Bug somewhere between cygwin, gcc and cvs
Hello all! I downloaded latest stable CVS code (1.11.2) and compiled it under cygwin. "cvs co", "cvs login" and "cvs commit" worked, but not "cvs update". I traced the bug to call to open (".", O_RDONLY); at lib/savecwd.c:61 in CVS source. This call is made the same way during commit and update, yet it fails during update only, with the error "file already exist". When I tried to make similar call from other places in the code, I learned that placing the call right before save_cwd is called from recurse.c makes the entire thing work. The next thing I tried was optimization level in the compiler. When I replaced -O2 with -O, the whole situation went away. I am running Windows ME. GCC gives the following in response to -v : $ gcc -v Reading specs from /usr/lib/gcc-lib/i686-pc-cygwin/3.2/specs Configured with: /netrel/src/gcc-3.2-3/configure --enable-languages=c,c++,f77,java --enable-libgcj --enable-threads=posix --with-system-zlib --enable-nls --without-included-gettext --enable-interpreter --disable-sjlj-exceptions --disable-version-specific-runtime-libs --enable-shared --build=i686-pc-linux --host=i686-pc-cygwin --target=i686-pc-cygwin --enable-haifa --prefix=/usr --exec-prefix=/usr --sysconfdir=/etc --libdir=/usr/lib --includedir=/nonexistent/include --libexecd ir=/usr/sbin Thread model: posix gcc version 3.2 20020927 (prerelease) cygwin version is cygwin 1.3.17-1 , according to cygcheck -- Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple Bug reporting: http://cygwin.com/bugs.html Documentation: http://cygwin.com/docs.html FAQ: http://cygwin.com/faq/