[M3commit] unresolve order TARGET vs. INSTALL_ROOT

[Jay K]

Yes and more so.

1) I don't know what the original full cause was when I ran makedist myself, to need unresolve fixes.
I just know unresolve/.m3ship had a problem that I didn't fully investigate, took a lazy route.
So those changes could be removed from release.


2) You could set stage to DARWIN_AMD64 or whatever instead of AMD64_DARWIN, working around this other part.
It depends how much you want -no-m3ship-resolve to work -- well enough for most paths and for us to release,
or also fix known problem.
I think -no-m3ship-resolve could be the one and only way -- always on, not an option -- once it works well enough.


3) You can take the second change without the first larger set as well.


4) *So far* we don't have a reason to go past RC5, except for very focused target-specific fixes, both for Darwin.
No portable bugs.
4b) But maybe that is wishful thinking.
4c) If we do have another release...nevermind...various thoughts about UtimeC.c...head has had
two restructurings since release..both of which would have fixed this problem, except I didn't know it.
The later restructuring caused source changes in callers, the earlier did not..
(64bit OpenBSD I think has the same problem as 64bit Darwin but so far we only have I386_OPENBSD
apparently releasing, and no NetBSD. Is there an official plan? Or, maybe part of the plan is, given
a tag, you can trickle lesser platforms out later? (My fault for not keeping VM hudson slaves up.)


 - Jay


----------------------------------------
> Date: Wed, 21 Apr 2010 17:11:53 +0200
> From: wagner at elegosoft.com
> To: jay.krell at cornell.edu
> CC: m3commit at elegosoft.com
> Subject: Re: [M3commit] unresolve order TARGET vs. INSTALL_ROOT
>
> Quoting Jay K :
>
>> This was already ahead of the tag, so not sure what to do. (I wish
>> we could add to the original mail subject lines..like take the first
>> line of the commit comment.)
>
> I think I understand it now.
> We only have the problem on AMD64_DARWIN because jbook2 is used to
> build two different targets, and I added AMD64_DARWIN to STAGE,
> which is the intermediate installation root. Weird.
>
> I can change STAGE for the time being and put the fix into the final
> release. Does that sound reasonable?
>
> Olaf
>
>
>> jbook2:cm3 jay$ cvs -z3 diff -u
>> cvs diff: Diffing .
>> cvs diff: Diffing src
>> Index: src/M3Build.m3
>> ===================================================================
>> RCS file: /usr/cvs/cm3/m3-sys/cm3/src/M3Build.m3,v
>> retrieving revision 1.28.2.11
>> diff -u -r1.28.2.11 M3Build.m3
>> --- src/M3Build.m3    17 Apr 2010 05:32:28 -0000    1.28.2.11
>> +++ src/M3Build.m3    21 Apr 2010 14:50:36 -0000
>> @@ -1899,8 +1899,17 @@
>>      res := TextUtils.Substitute(res, t.emacs_install_alt, "\" &
>> EMACS_INSTALL & \"");
>>      res := TextUtils.Substitute(res, t.pkg_install_alt, "\" &
>> PKG_INSTALL & \"");
>>      res := TextUtils.Substitute(res, t.pkg_use_alt, "\" & PKG_USE & \"");
>> -    res := TextUtils.Substitute(res, t.text_build_dir, "\" & TARGET & \"");
>> +
>> +    (* INSTALL_ROOT needs to be relatively late since it is a
>> prefix of others.
>> +     * TARGET needs to be even later since it can occur arbitrarily, as in:
>> +     * e.g. /Users/hudson/workspace/makedist-AMD64_DARWIN/bin
>> +     *                                       ^--TARGET--^
>> +     *      ^--------------------------- INSTALL_ROOT --^
>> +     *      ^--------------------------- BIN_INSTALL  ------^
>> +     *)
>>      res := TextUtils.Substitute(res, t.install_root_alt, "\" &
>> INSTALL_ROOT & \"");
>> +    res := TextUtils.Substitute(res, t.text_build_dir, "\" & TARGET & \"");
>> +
>>      res := TextUtils.Substitute(res, "\"\" & ", "");
>>      res := TextUtils.Substitute(res, " & \"\"", "");
>>      RETURN res;
>>
>>  - Jay
>>
>>
>> ----------------------------------------
>>> Date: Wed, 21 Apr 2010 16:52:33 +0000
>>> To: m3commit at elegosoft.com
>>> From: jkrell at elego.de
>>> Subject: [M3commit] CVS Update: cm3
>>>
>>> CVSROOT: /usr/cvs
>>> Changes by: jkrell at birch. 10/04/21 16:52:33
>>>
>>> Modified files:
>>> cm3/m3-sys/cm3/src/: Tag: release_branch_cm3_5_8 M3Build.m3
>>>
>>> Log message:
>>> TARGET must be back-substituted later, after INSTALL_ROOT, Olaf
>>> correctly points out.
>>>
>>
>
>
>
> --
> Olaf Wagner -- elego Software Solutions GmbH
> Gustav-Meyer-Allee 25 / Gebäude 12, 13355 Berlin, Germany
> phone: +49 30 23 45 86 96 mobile: +49 177 2345 869 fax: +49 30 23 45 86 95
> http://www.elegosoft.com | Geschäftsführer: Olaf Wagner | Sitz: Berlin
> Handelregister: Amtsgericht Charlottenburg HRB 77719 | USt-IdNr: DE163214194
>