|
From: R. B. <ro...@pa...> - 2007-02-27 04:57:35
|
In preparation for the first Debian packaging, I'd like to put out another release of remake soon. The changes are mostly small: * Changes to make more Debian compatibile. * Track some GNU Make 3.81 changes, which include a bugfix and making more strings "const". * Misc bugfixes If you can try out what's in CVS, I'd appreciate. If you want me to put out a candidate tarball, let me know. Thanks |
|
From: Dave K. <dav...@ar...> - 2007-03-06 16:25:03
Attachments:
remake-autogen.log
|
On 27 February 2007 04:58, R. Bernstein wrote:
> If you can try out what's in CVS, I'd appreciate. If you want me to
> put out a candidate tarball, let me know. Thanks
I checked out CVS HEAD, and autogen.sh fails in multiple interesting ways.
Could be an auto* version mismatch? I'm using am 1.10 and ac 2.60.
cheers,
DaveK
--
Can't think of a witty .sigline today....
|
|
From: R. B. <ro...@pa...> - 2007-03-07 03:47:32
|
Dave Korn writes: > On 27 February 2007 04:58, R. Bernstein wrote: > > > > If you can try out what's in CVS, I'd appreciate. If you want me to > > put out a candidate tarball, let me know. Thanks > > > I checked out CVS HEAD, and autogen.sh fails in multiple interesting ways. > Could be an auto* version mismatch? I'm using am 1.10 and ac 2.60. I've been using automake 1.9.6 and autoconf 2.59, 2.60 and 2.61. Alas going to take be a while before I can get that combination of 1.10 and 2.60. What are the error messages? Thanks. > -- > Can't think of a witty .sigline today.... I never can. |
|
From: Dave K. <dav...@ar...> - 2007-03-07 10:23:43
|
On 07 March 2007 03:47, R. Bernstein wrote:
> Dave Korn writes:
> > On 27 February 2007 04:58, R. Bernstein wrote:
> >
> >
> > > If you can try out what's in CVS, I'd appreciate. If you want me to
> > > put out a candidate tarball, let me know. Thanks
> >
> >
> > I checked out CVS HEAD, and autogen.sh fails in multiple interesting
> ways. > Could be an auto* version mismatch? I'm using am 1.10 and ac 2.60.
>
> I've been using automake 1.9.6 and autoconf 2.59, 2.60 and 2.61. Alas
> going to take be a while before I can get that combination of 1.10 and
> 2.60.
Right, fair enough; Cygwin supports switchable auto* versions, so I'll have
another go with 1.9 and let you know.
>
> What are the error messages?
There was an attachment on that last post, did you receive it correctly?
cheers,
DaveK
--
Can't think of a witty .sigline today....
|
|
From: R. B. <ro...@pa...> - 2007-03-07 13:00:33
|
My mistake - you are correct, there was a log attached. And yes those *are* interesting new messages. I've committed a change to reduce some of these. I will probably do more later and when I get a computer that can have automake 1.10 installed. So it's now an error to call AC_PROC_CC after AM_INIT_AUTOMAKE (or is it AM_CONFIG_HEADER)? Fine, I'll remove it. And the warning about "immediate" assigments (:=) are a lame attempt to get the make process to speed up. But they've caused programs in BSD's make and even a bug. So out they go. I'm always amazed at how much time I spend feeding and attending to the automake system. In fact that's why remake was undertaken. Yes, that Dave Korn reports: ... > > There was an attachment on that last post, did you receive it correctly? > Can't think of a witty .sigline today.... Not much of a fan of "fortune" either. |
|
From: Dave 'N. n. t. one!' K. <dav...@ar...> - 2007-03-07 14:47:26
|
On 07 March 2007 13:01, R. Bernstein wrote:
> My mistake - you are correct, there was a log attached. And yes those
> *are* interesting new messages. I've committed a change to reduce some
> of these. I will probably do more later and when I get a computer that
> can have automake 1.10 installed.
Switching to automake-1.9.6 fixed all the problems and allowed autogen.sh to
complete successfully. After updating from CVS, 1.10 still doesn't succeed,
but it's less noisy with the changes you've added. If I get some spare time
to go into it in detail, I'll send patches for any problems I can fix, but
auto* is not entirely my forte.
> I'm always amazed at how much time I spend feeding and attending to
> the automake system. In fact that's why remake was undertaken.
It's also just the /only/ way to debug really complex makefiles where you've
got loads of rules and dependencies being calculated and eval'd at runtime.
It's also a very useful learning tool. Stepping through makefiles makes it
sooo much easier to understand what's going on and what's happening with
complex function calls and multi-line definitions. Using it has allowed me to
greatly extend my understanding of advanced makefile techniques. Thank you
very much.
> Yes, that Dave Korn reports:
^^^^^^^^^^^^^^^^^^^
;-) I refer you to my current 'From:' line...
cheers,
DaveK
--
I have discovered a truly magnificently witty .sigline
|
|
From: R. B. <ro...@pa...> - 2007-03-09 02:05:31
|
> Switching to automake-1.9.6 fixed all the problems and allowed > autogen.sh to > complete successfully. After updating from CVS, 1.10 still doesn't > succeed, > but it's less noisy with the changes you've added. If I get some spare > time > to go into it in detail, I'll send patches for any problems I can fix, but > auto* is not entirely my forte. The fact that on FreeBSD (and other OS's) I see 4 versions of automake and 4 versions autoconf suggest these folks have a real tough time with auto* compatibility. I looked at some of the other warning and error messages from automake. automake kibbutzes that I should use things, like AM_PATH_LISPDIR which I do use. Or maybe I use those differently. Yet, when I go to the automake manual (and here I've always found it easier and more thorough to edit the texinfo source than try to navigate inside info), I don't see a single example of how one is supposed to use AM_PATH_LISPDIR. The only book I know of on this topic "GNU Autoconf, automake and libtool" published by New Riders also is silent on this, and given what was written above and that was published in 2000, it's probably totally out of date. > >> I'm always amazed at how much time I spend feeding and attending to >> the automake system. In fact that's why remake was undertaken. > > It's also just the /only/ way to debug really complex makefiles where > you've > got loads of rules and dependencies being calculated and eval'd at > runtime. > > It's also a very useful learning tool. Stepping through makefiles makes > it > sooo much easier to understand what's going on and what's happening with > complex function calls and multi-line definitions. Using it has allowed > me to > greatly extend my understanding of advanced makefile techniques. Thank > you > very much. > >> Yes, that Dave Korn reports: > ^^^^^^^^^^^^^^^^^^^ > > ;-) I refer you to my current 'From:' line... > > cheers, > DaveK > -- > I have discovered a truly magnificently witty .sigline -- This is not a .sigline |