SAGE_SPKG_CONFIGURE macro: Add new pre-check and post-check optional arguments

[embray]

This adds two new optional arguments to the SAGE_SPKG_CONFIGURE. Both are sets of actions that should be run unconditionally when running the resulting configure script, regardless of the outcome of checks for the package.

This demonstrates a use case for this in mpir, where no matter what else it is imperative to set a value for SAGE_MP_LIBRARY (even if we are not installing the mpir SPKG, or the SPKG is already installed).

The follow-up ticket #27642 might make more clear why this is useful: If some SPKG is already installed we shouldn't bother checking for it on the system, so we skip the dependency checks. But we may still want to perform other package-specific configure actions.

One thing I don't like about this entirely is that it would seem clearer and more logical to make the "pre" argument come before the other arguments. However, since it's optional, and is not used for many packages, I thought it would be better to add it to the end of the argument list and not have to update every spkg-configure.m4 to pass an empty value for the first argument.

CC: @dimpase

Component: build: configure

Author: Erik Bray

Branch/Commit: 1c24df1

Reviewer: Dima Pasechnik

Issue created by migration from https://trac.sagemath.org/ticket/27641

[embray]

Description changed:

--- 
+++ 
@@ -2,6 +2,6 @@
 
 This demonstrates a use case for this in mpir, where no matter what else it is imperative to set a value for `SAGE_MP_LIBRARY` (even if we are not installing the mpir SPKG, or the SPKG is already installed).
 
-It will maybe be more clear why this is needed in a follow-up I will add next.
+The follow-up ticket #27642 might make more clear why this is useful: If some SPKG is already installed we shouldn't bother checking for it on the system, so we skip the dependency checks.  But we may still want to perform other package-specific configure actions.
 
 One thing I don't like about this entirely is that it would seem clearer and more logical to make the "pre" argument come before the other arguments.  However, since it's optional, and is not used for many packages, I thought it would be better to add it to the end of the argument list and not have to update every `spkg-configure.m4` to pass an empty value for the first argument.

[dimpase]

comment:3

This seems to depend, and use the branch, of #27212, no?

[dimpase]

Changed branch from u/embray/build/configure/pre-post to public/build/configure/pre-post

[dimpase]

Changed commit from a2bab81 to f9c9a91

[dimpase]

comment:4

rebased over 8.8.beta3

---

New commits:

191269e	Add support for unconditional pre- and post-check actions in spkg-configure.m4
f9c9a91	Refactor the mpir/spkg-configure.m4 (and to a lesser extent the one for gmp) to use the new pre and post arguments to SAGE_SPKG_CONFIGURE

[embray]

comment:5

That's a good point; yes it included #27212. Thanks for rebasing.

[dimpase]

comment:6

Do you have an example use of the macro where the 3rd argument is not empty?
(Perhaps from one of the tickets on the list #27330 ?)

And, by the way, docs explaining them are hard to read, could you please instead write:
"1st argument does this and this", "2nd argument does...", etc?

[dimpase]

comment:7

It looks as if it basically adds a bit of structure into spkg-configure.m4 files (i.e. logically different parts are inside []), no more than that, right? Can this wait until it gets used for actual refactoring?

[embray]

comment:8

Replying to @dimpase:

Do you have an example use of the macro where the 3rd argument is not empty?
(Perhaps from one of the tickets on the list #27330 ?)

The (currently) third argument is used in a few packages. At least one example I can think of off the top of my head is yasm. The point of the third argument is to check whether that dependency is needed at all for anything, as opposed to the second argument which assumes the package is required but does the check for whether it's provided already by the system.

And, by the way, docs explaining them are hard to read, could you please instead write:
"1st argument does this and this", "2nd argument does...", etc?

I can try to reword it.

[embray]

comment:9

Replying to @dimpase:

It looks as if it basically adds a bit of structure into spkg-configure.m4 files (i.e. logically different parts are inside []), no more than that, right?

Essentially, yes. It just allows insertion of commands into a couple places where it wasn't previously possible to insert them.

Can this wait until it gets used for actual refactoring?

I'm not sure what you mean here. It is used for #27642 which has this ticket as a dependency.