[Fwd: Announce: corbaconf-1.0.1 is out]
corbaconf is a autoconf based package, which support building multiplatform CORBA applications. 1.0.1 version is released today and aviable via http://corbaconf.kiev.ua Regards.
Re: grep -E ? (Was: Compilation problem solved)
Hello, Mathieu! I have tried the same code with egcs-2.91.66 and the errors (GCC_* and families missing) are not reported. So the problem come from gcc. gcc -v reports: 2.96 2724 (experimental). Oops, I didn't see the experimental thing... The difference is in the -Wundef flag given to the preprocessor (gcc -E). Without it, the results are the same between the two gcc. The strange thing, is that the gcc info page tells: `-Wundef' Warn if an undefined identifier is evaluated in an `#if' directive. My guess is that the experimental version of GCC you are using is less forgiving to the system headers, at least on the open-source systems. gcc-2.95.2 and the older versions would not apply -Wundef to the system headers. This essentially means that if you want to picky to the code you are writing or testing (using GRUB from CVS implies testing it) and you are using GNU/Linux (I hope that GCC is more forgiving on commercial Unices) you should also be picky to the headers installed on your system, i.e. report the problem to the ncurses maintainers. However, I'm still curious how this warning (not error) drove the configure script into believing that ncurses in not installed at all. I'll appreciate if you send me config.log privately. Regards, Pavel Roskin
AC_TRY_CPP too strict? (Was: grep -E)
Hello again! However, I'm still curious how this warning (not error) drove the configure script into believing that ncurses in not installed at all. I'll appreciate if you send me config.log privately. Sorry, I posted my reply too early. No need to send me anything. It's obvious from the definition of AC_TRY_CPP that warnings in headers are not tolerated. This can be reproduced by this configure.in: AC_INIT AC_CHECK_HEADERS(ncurses/curses.h) AC_OUTPUT $ autoconf $ CC="gcc -Wundef -I../../../../../usr/include" ./configure checking how to run the C preprocessor... gcc -Wundef -I../../../../../usr/include -E checking for ncurses/curses.h... no creating ./config.status It remains an open question whether Autoconf should be so strict. Regards, Pavel Roskin
Re: AC_TRY_CPP too strict? (Was: grep -E)
On Aug 31, 2000, Pavel Roskin [EMAIL PROTECTED] wrote:
It's obvious from the definition of AC_TRY_CPP that warnings in
headers are not tolerated.
It remains an open question whether Autoconf should be so strict.
It must. There are compilers that don't return a failure status when
asked to #include a non-existent file.
--
Alexandre Oliva Enjoy Guarana', see http://www.ic.unicamp.br/~oliva/
Red Hat GCC Developer aoliva@{cygnus.com, redhat.com}
CS PhD student at IC-Unicampoliva@{lsd.ic.unicamp.br, gnu.org}
Free Software Evangelist*Please* write to mailing lists, not to me
Re: HTML format documentation
Richard Stallman writes: If prefix is `/usr', use `/etc'. Otherwise, use $(prefix)/config. I hope not. Hard-coded exceptions are the path to insanity. :) So prefix=/usr leads to syconfidir=/etc, but prefix=/usr/gnu (not my idea) leads to sysconfdir=/usr/gnu/config? Ugh. Or perhaps the "Linux" folks will start arguing that prefix=/usr should imply localstatedir=/var. Or maybe prefix=/usr should imply prefix=/? The nice thing about the GNU file system is that it has no exceptions. -- Peter Eisentraut Sernanders väg 10:115 [EMAIL PROTECTED] 75262 Uppsala http://yi.org/peter-e/Sweden
Web archives of this mailing list?
Can anyone tell me if there are web archives of this mailing list available? If not, can I retrieve archives via majordomo commands? cgf
Re: Web archives of this mailing list?
Hello, Chris! On Thu, 31 Aug 2000, Chris Faylor wrote: Can anyone tell me if there are web archives of this mailing list available? If not, can I retrieve archives via majordomo commands? Mailing lists archived on sources.redhat.com: http://sources.redhat.com/lists.html I'm not sure about majordomo. Send "help" to [EMAIL PROTECTED] if you want to find out. Regards, Pavel Roskin
Re: Web archives of this mailing list?
Chris Can anyone tell me if there are web archives of this mailing Chris list available? If not, can I retrieve archives via majordomo Chris commands? I believe the archives are available on sources.redhat.com. Are there plans to move the archives to an FSF server? What about the Gnats database? Tom
Re: Web archives of this mailing list?
When did the autoconf list start setting Reply-To? This is very lame. Tom
Re: Web archives of this mailing list?
On Aug 31, 2000, Chris Faylor [EMAIL PROTECTED] wrote:
Can anyone tell me if there are web archives of this mailing list
available?
ftp://ftp-mailing-list-archive.gnu.org/current/autoconf
http://www.cygnus.com/ml/autoconf/
If not, can I retrieve archives via majordomo commands?
I don't think so; it's not majordomo that runs GNU lists.
--
Alexandre Oliva Enjoy Guarana', see http://www.ic.unicamp.br/~oliva/
Red Hat GCC Developer aoliva@{cygnus.com, redhat.com}
CS PhD student at IC-Unicampoliva@{lsd.ic.unicamp.br, gnu.org}
Free Software Evangelist*Please* write to mailing lists, not to me
Re: grep -E ? (Was: Compilation problem solved)
Hello, Alexandre! `+' isn't a portable literal. Some implementations seem to use it as a meta-character with the usual meaning, but without support for it after `*'. Alexandre, could you be more exact? No. I seem to recall some comments about non-portability of `+', and certainly associated them with this bug report when it came up. In other words, don't pay attention to what I wrote. If Paul said it's ok to assume `+' is not a meta-character in `grep', I believe it. You shouldn't believe it. The answer is Plan 9. See the discussion in [EMAIL PROTECTED] starting with http://sources.redhat.com/ml/autoconf-prs/2000-q3/msg00015.html This means that my patch for enclosing plus in square backets should be applied, maybe with a comment about Plan 9. PR/138 should be closed after that. Anybody? Regards, Pavel Roskin
Re: Web archives of this mailing list?
On Thu, Aug 31, 2000 at 02:32:44PM -0400, Pavel Roskin wrote: On Thu, 31 Aug 2000, Chris Faylor wrote: Can anyone tell me if there are web archives of this mailing list available? If not, can I retrieve archives via majordomo commands? Mailing lists archived on sources.redhat.com: http://sources.redhat.com/lists.html I'm not sure about majordomo. Send "help" to [EMAIL PROTECTED] if you want to find out. Duh. I completely missed that this was hosted on sources.redhat.com. That's particularly embarrassing since I'm one of the sysadmins for that machine. :-) Thanks for the info. cgf
Re: grep -E ? (Was: Compilation problem solved)
Date: Thu, 31 Aug 2000 14:41:48 -0400 (EDT) From: Pavel Roskin [EMAIL PROTECTED] If Paul said it's ok to assume `+' is not a meta-character in `grep', I believe it. You shouldn't believe it. The answer is Plan 9. OK, but Plan 9 grep is not the same as traditional grep at all. It uses a syntax that is almost an extension of POSIX egrep (i.e. it uses EREs, not BREs). Unfortunately it's not a pure extension: the bracket expression syntax is incompatible. If you want autoconf to work with Plan 9, you'll probably have more work to do. For example, Plan 9 does not have "egrep", so you'll have to remove all uses of egrep (despite what the GNU coding standards say). And there are undoubtedly other problems as well. Can you ask Plan 9 installers to use the Plan 9 ANSI-POSIX Environment (APE) instead? http://plan9.bell-labs.com/sys/doc/ape.html claims that they have a POSIX grep.
Re: grep -E ? (Was: Compilation problem solved)
Date: Thu, 31 Aug 2000 14:41:48 -0400 (EDT) From: Pavel Roskin [EMAIL PROTECTED] If Paul said it's ok to assume `+' is not a meta-character in `grep', I believe it. You shouldn't believe it. The answer is Plan 9. OK, that's news to me. But you should be careful: Plan 9 grep is not the same as traditional grep at all. It uses a syntax that is almost an extension of POSIX egrep (i.e. it uses EREs, not BREs). Unfortunately the syntax is not a pure extension of EREs, as the bracket expression syntax is incompatible. If you want autoconf to work with Plan 9, you'll probably have more work to do. For example, the documentation says that Plan 9 does not have "egrep", so you'll have to remove all uses of egrep (despite what the GNU coding standards say). And there are undoubtedly other problems as well. Can you ask Plan 9 installers to use the Plan 9 ANSI-POSIX Environment (APE) instead? http://plan9.bell-labs.com/sys/doc/ape.html claims that they have a POSIX grep. That might save you some hassles.
Re: Web archives of this mailing list?
On Aug 31, 2000, Tom Tromey [EMAIL PROTECTED] wrote:
When did the autoconf list start setting Reply-To?
Yours was the first message with a Reply-To. Let's see if it happens
to me too...
This is very lame.
Indeed.
--
Alexandre Oliva Enjoy Guarana', see http://www.ic.unicamp.br/~oliva/
Red Hat GCC Developer aoliva@{cygnus.com, redhat.com}
CS PhD student at IC-Unicampoliva@{lsd.ic.unicamp.br, gnu.org}
Free Software Evangelist*Please* write to mailing lists, not to me
Re: grep -E ? (Was: Compilation problem solved)
If Paul said it's ok to assume `+' is not a meta-character in `grep', I believe it. You shouldn't believe it. The answer is Plan 9. OK, but Plan 9 grep is not the same as traditional grep at all. It uses a syntax that is almost an extension of POSIX egrep (i.e. it uses EREs, not BREs). Unfortunately it's not a pure extension: the bracket expression syntax is incompatible. If you want autoconf to work with Plan 9, you'll probably have more work to do. For example, Plan 9 does not have "egrep", so you'll have to remove all uses of egrep (despite what the GNU coding standards say). And there are undoubtedly other problems as well. I'm affraid that this subtle problem starts getting more attention that is deserves. My arguments (if neighter of them is convincing, let's forget this issue and switch to something more interesting): 1) The lack of egrep is easier to detect than this quirk in grep. 2) It is Ok to ask for additional programs, but it's better to avoid replacing the programs the system ships with. 3) There is an open PR/138 in the autoconf bug database. Somebody must have encountered this already. 4) Somebody has already been confused by an unquoted plus while having problems with ncurses headers. Regards, Pavel Roskin
Re: AC_TRY_CPP too strict? (Was: grep -E)
Hello, Alexandre! It remains an open question whether Autoconf should be so strict. It must. There are compilers that don't return a failure status when asked to #include a non-existent file. I propose a new item to the TODO list (for after 2.50): Check if the compiler returns a failure status when asked to #include a non-existent file. If it does, don't check stderr from cpp, only check the exit status. There are hundreds things a compiler can complain about. An example that I remember - the native compiler on HP-UX would warn that the code it generates doesn't run on earlier processors. If the compiler is "wary" and indicates missing includes by the exit status its subtle warnings should be acceptable. If the compiler is "careless" its warnings should be taken with care. Also it is better for developers to show warnings while compiling the code than to hide them into config.log Regards, Pavel Roskin
Re: subversions.gnu.org?
On Tue, Aug 29, 2000 at 07:54:02PM -0400, Harlan Stenn wrote: Is subversions OK? I went to "cvs update" my autoconf repository and couldn't access it. It was poorly for a day or two. I couldn't access it at all yesterday, but was able to do some check ins this evening. Cheers, Gary. -- ___ _ ___ __ _ mailto: [EMAIL PROTECTED] / __|__ _ _ ___ _| | / / | / /_ _ _ _ __ _| |_ __ _ ___ [EMAIL PROTECTED] | (_ / _` | '_|// / |/ /| |/ / _` | || / _` | ' \/ _` | _ \ \___\__,_|_|\_, /|___(_)___/\__,_|\_,_\__, |_||_\__,_|//_/ home page: /___/ /___/ gpg public key: http://www.oranda.demon.co.uk http://www.oranda.demon.co.uk/key.asc
Re: AC_LIBOBJ
On Thu, Aug 31, 2000 at 01:45:30AM -0300, Alexandre Oliva wrote: On Aug 29, 2000, Harlan Stenn [EMAIL PROTECTED] wrote: The docs for AC_LIBOBJ say (in part): Technically, it adds 'FUNCTION.$ac_objext' to ... If this is true, it may still be a bug. Perhaps it should add: 'FUNCTION$U.$ac_objext' because some folks will be using the ansi2knr stuff. Isn't it automake that adjusts LIBOBJS to work with ansi2knr? Nope. The user has to add a sed script snippet to their configure.in if they use LIBOBJS in conjunction with either LTLIBRARIES or ansi2knr. According to the excellent `Autoconf, Automake and Libtool' [New Riders Publishing; ISBN: 1578701902 =)O| =)O| ]: # This is necessary so that .o files in LIBOBJS are also built via # the ANSI2KNR-filtering rules. Xsed='sed -e "s/^X//"' LIBOBJS=`echo X"$LIBOBJS"|\ $Xsed 's/\.[^.]* /.\$U /g;s/\.[^.]*$/.\$U/'` # LTLIBOBJS requires objects with .lo suffixes. LTLIBOBJS=echo X"$LIBOBJS" | \ $Xsed 's/\.[^.]* /.lo /g;s/\.[^.]*$/.lo/' AC_SUBST(LTLIBOBJS) Cheers, Gary. -- ___ _ ___ __ _ mailto: [EMAIL PROTECTED] / __|__ _ _ ___ _| | / / | / /_ _ _ _ __ _| |_ __ _ ___ [EMAIL PROTECTED] | (_ / _` | '_|// / |/ /| |/ / _` | || / _` | ' \/ _` | _ \ \___\__,_|_|\_, /|___(_)___/\__,_|\_,_\__, |_||_\__,_|//_/ home page: /___/ /___/ gpg public key: http://www.oranda.demon.co.uk http://www.oranda.demon.co.uk/key.asc
Re: HTML format documentation
Here is an idea. Suppose that the default for sysconfdir were computed from the actual value of prefix, as follows: If prefix is `/usr', use `/etc'. Otherwise, use $(prefix)/config. I think we need to ask the community of developers what they would think of this change. It sounds good, but it could have pitfalls we have not thought of. Would someone like to do that?
Re: grep -E ? (Was: Compilation problem solved)
On Thu, Aug 31, 2000 at 03:50:48PM -0400, Pavel Roskin wrote: 4) Somebody has already been confused by an unquoted plus while having problems with ncurses headers. ...but we never did find the root cause -- but found some suggested workarounds that only lead into further discussion. -- Thomas E. Dickey [EMAIL PROTECTED] http://dickey.his.com ftp://dickey.his.com
Re: grep -E ? (Was: Compilation problem solved)
On Thu, Aug 31, 2000 at 12:19:56PM -0700, Paul Eggert wrote: OK, but Plan 9 grep is not the same as traditional grep at all. It uses a syntax that is almost an extension of POSIX egrep (i.e. it uses EREs, not BREs). Unfortunately it's not a pure extension: the bracket expression syntax is incompatible. in short: "improved beyond all recognition" -- Thomas E. Dickey [EMAIL PROTECTED] http://dickey.his.com ftp://dickey.his.com
C++ Shared Libraries
From what I have read about libtool it cannot handle C++ shared libs. Is there any work going on in this area, or should I just do my best with shell scripting in configure.in? Anyone have a better answer? -- Robert Boehne Software Engineer Ricardo Software Chicago Technical Center TEL: (630)789-0003 x. 238 FAX: (630)789-0127 email: [EMAIL PROTECTED]
Re: AC_LIBOBJ
On Thu, Aug 31, 2000 at 02:34:41PM -0700, Bruce Korb wrote: "Gary V. Vaughan" wrote: According to the excellent `Autoconf, Automake and Libtool' [New Riders Publishing; ISBN: 1578701902 =)O| =)O| ]: # This is necessary so that .o files in LIBOBJS are also built via # the ANSI2KNR-filtering rules. Xsed='sed -e "s/^X//"' LIBOBJS=`echo X"$LIBOBJS"|\ $Xsed 's/\.[^.]* /.\$U /g;s/\.[^.]*$/.\$U/'` # LTLIBOBJS requires objects with .lo suffixes. LTLIBOBJS=echo X"$LIBOBJS" | \ $Xsed 's/\.[^.]* /.lo /g;s/\.[^.]*$/.lo/' AC_SUBST(LTLIBOBJS) I *do* hope this example is not in the book :-( :-(. Your Xsed macro needs to be written: Xsed='sed -e "s/^X//" -e ' or your usage of it needs to be changed. Mistranscribed. The $Xsed's are indeed followed by a `-e'. Cheers, Gary. -- ___ _ ___ __ _ mailto: [EMAIL PROTECTED] / __|__ _ _ ___ _| | / / | / /_ _ _ _ __ _| |_ __ _ ___ [EMAIL PROTECTED] | (_ / _` | '_|// / |/ /| |/ / _` | || / _` | ' \/ _` | _ \ \___\__,_|_|\_, /|___(_)___/\__,_|\_,_\__, |_||_\__,_|//_/ home page: /___/ /___/ gpg public key: http://www.oranda.demon.co.uk http://www.oranda.demon.co.uk/key.asc
