configure.ac: write build/make/Makefile within an AC_CONFIG_FILE, not during main configure

[mkoeppe]

build/make/Makefile should be written by config.status within an AC_CONFIG_FILE template, not at configure time.

Depends on #24729 Depends on #24703

Component: build: configure

Author: Erik Bray

Branch: 2a99185

Reviewer: Julian Rüth

Issue created by migration from https://trac.sagemath.org/ticket/21524

[embray]

comment:41

Replying to @jdemeyer:

Replying to @embray:

I sympathize with your concerns but I strongly suspect that they will prove overstated.

I really hope that you are right and that this won't turn in a "I told you so" scenario.

If it does, I promise that I'll graciously accept that you were right :)

[embray]

comment:42

Also, I think that I've addressed most of the other review comments so far....

[7ed8c4ca-6d56-4ae9-953a-41e42b4ed313]

Changed commit from 7668fcd to 735fd36

[7ed8c4ca-6d56-4ae9-953a-41e42b4ed313]

Branch pushed to git repo; I updated commit sha1. This was a forced push. New commits:

4b4f40e	Add --with-python=3 configure flag to replace SAGE_PYTHON3=yes
0b36d3e	Use AC_CONFIG_FILES + AC_OUTPUT to generate build/make/Makefile
e86ea4c	Further rewrite of configure.ac to do away with the multiple
586eb4a	Remove some bogus comments
1a04d71	Improved (?) documentation of the Makefile.in template.
4dbf15b	Get SHELL from autoconf
7cddac9	Delete redundant AC_CONFIG_MACRO_DIR
d889070	Added a little bit more information about the build system and how it works, at a high level
735fd36	Fix sage -i

[embray]

comment:44

Rebased to incorporate #24729

[embray]

Dependencies: #24729

[7ed8c4ca-6d56-4ae9-953a-41e42b4ed313]

Branch pushed to git repo; I updated commit sha1. This was a forced push. New commits:

057b434	Everything should be rebuilt after GCC upgrade
4056a0d	Merge branch 'u/jdemeyer/everything_should_be_rebuilt_after_gcc_upgrade' into HEAD
025804c	Use AC_CONFIG_FILES + AC_OUTPUT to generate build/make/Makefile
83bc9fc	Further rewrite of configure.ac to do away with the multiple
87c7389	Remove some bogus comments
283ec7b	Improved (?) documentation of the Makefile.in template.
3b512dd	Get SHELL from autoconf
7a9fff3	Delete redundant AC_CONFIG_MACRO_DIR
869f57a	Added a little bit more information about the build system and how it works, at a high level
1c02a0e	Fix sage -i

[7ed8c4ca-6d56-4ae9-953a-41e42b4ed313]

Changed commit from 735fd36 to 1c02a0e

[embray]

comment:46

Rebased again and incorporated #24703.

[embray]

Changed dependencies from #24729 to #24729, #24703

[fchapoton]

comment:47

needs rebase

[7ed8c4ca-6d56-4ae9-953a-41e42b4ed313]

Branch pushed to git repo; I updated commit sha1. This was a forced push. New commits:

5b94974	Everything should be rebuilt after GCC upgrade
54f3a48	Use AC_CONFIG_FILES + AC_OUTPUT to generate build/make/Makefile
bdeca46	Further rewrite of configure.ac to do away with the multiple
1150f95	Remove some bogus comments
505db02	Improved (?) documentation of the Makefile.in template.
c61dba4	Get SHELL from autoconf
eefce2c	Delete redundant AC_CONFIG_MACRO_DIR
d77e687	Added a little bit more information about the build system and how it works, at a high level
6f99aab	Fix sage -i

[7ed8c4ca-6d56-4ae9-953a-41e42b4ed313]

Changed commit from 1c02a0e to 6f99aab

[embray]

comment:49

The longer this stays open the harder it is to keep it maintained because it will conflict with anything that modifies configure.ac.

[saraedum]

comment:50

Ok. Reviewing this now…

[saraedum]

Changed work issues from minor documentation issues to none

[saraedum]

comment:51

Thanks for this very nice summary in the README. If you have checked that this actually works, feel free to set this to positive review.

[embray]

comment:52

There's one last change I would like to make, though it shouldn't substantially alter anything. I just want to move all the stuff that collects the packages into its own macro, and move it to a separate file.

I already did this in another branch, but it doesn't make particular sense to leave out of this ticket.

[7ed8c4ca-6d56-4ae9-953a-41e42b4ed313]

Changed commit from 6f99aab to bb9ff43

[7ed8c4ca-6d56-4ae9-953a-41e42b4ed313]

Branch pushed to git repo; I updated commit sha1. New commits:

bb9ff43

Moved the code that collects up SPKG info into a single SAGE_SPKG_COLLECT macro, along with additional documentation thereof.

[embray]

comment:54

Will be interesting to see how the buildbots do with this.

[embray]

comment:55

There appears to be a bug in outputting the package versions during the configure checks. This might have been introduced in my last rebase...

[7ed8c4ca-6d56-4ae9-953a-41e42b4ed313]

Branch pushed to git repo; I updated commit sha1. New commits:

2a99185

This should be quoted to prevent the $1 from being expanded as an argument to the m4 macro

[7ed8c4ca-6d56-4ae9-953a-41e42b4ed313]

Changed commit from bb9ff43 to 2a99185

[saraedum]

comment:58

Ok. Your recent changes look good to me.

[vbraun]

Changed branch from u/embray/build/makefile-in to 2a99185

[jhpalmieri]

Changed commit from 2a99185 to none

[jhpalmieri]

comment:60

This completely breaks upgrading Sage, it seems to me. On three different OS X machines with 8.2.beta7 installed, upgrading to 8.2.beta8, or in particular just using this branch, results in a bad build/make/Makefile: lines like

deps_4ti2 =  $(SAGE_LOCAL)/bin/gcczlib $(MP_LIBRARY) glpk
deps_alabaster =  $(SAGE_LOCAL)/bin/gcc$(PYTHON) | pip
deps_appnope =  $(SAGE_LOCAL)/bin/gcc$(PYTHON) | pip

Please fix this.

[embray]

comment:61

Indeed, it looks like a bug introduced upon the last rebase. I didn't catch it because it only affects if you are using Sage's gcc spkg.

[embray]

comment:62

Actually it appears the problem might come from #24703 in the first place, since it seems to misplace the space in the GCC_DEP variable...

[embray]

comment:63

Replying to @embray:

Actually it appears the problem might come from #24703 in the first place, since it seems to misplace the space in the GCC_DEP variable...

I take that back--the missing space is in my code.

[jdemeyer]

comment:65

You cannot just re-open a ticket. Only the release manager should do that.

[jdemeyer]

comment:66

Especially this ticket must not be re-opened since it has already been released. Just open a new ticket to fix it.

[embray]

comment:67

It's not in the latest beta yet.

[embray]

comment:68

Oops, apparently it is. If I had thought otherwise I wouldn't have reopened. Strange--I just checked too any I didn't see it.

[embray]

comment:69

Followup in #24961.

[jhpalmieri]

comment:70

Thanks for the quick diagnosis, by the way!

[embray]

comment:71

Replying to @jhpalmieri:

Thanks for the quick diagnosis, by the way!

No, thank you for pointing it out.

[jdemeyer]

comment:72

Why is this making changes to build/pkgs/gcc/spkg-install? That might have broken building Sage from scratch if GCC is built.

[jdemeyer]

comment:73

This might have broken $(SAGERUNTIME): #24995

[embray]

comment:74

Replying to @jdemeyer:

Why is this making changes to build/pkgs/gcc/spkg-install? That might have broken building Sage from scratch if GCC is built.

I just noticed this comment. That was not intentional. Probably a mis-resolved conflict (though I don't know why since I never made any edits to that file in the first place).

[jdemeyer]

comment:75

DUMMY_PACKAGES seems broken: #25188

[jdemeyer]

comment:76

What was the purpose of this change:

@@ -58,11 +58,11 @@ all-sage: \
 # option to make forces all targets to be built unconditionally)
 download-for-sdist: base
        env SAGE_INSTALL_FETCH_ONLY=yes $(MAKE) -B SAGERUNTIME= \
-               $(SDIST_PACKAGES)
+               $(SDIST_PACKAGE_INSTS)

It has broken sdists, see #25319.