Viktor:
>Please be specific: What did I remove which broke 3.76?
I explained since first message and perhaps is necessary to add info
First problem with make376 is with "inner dir separators"
Before we fixed it with expressions like:
OS2_LIB_ARCH = $(subst /,\\,$(LIB_ARCH))
-CMD.EXE /C IF EXIST $(OS2_ARCH_DIR)\*.* $(RM) $(OS2_ARCH_DIR)\\*.*
which no longer exist ( note \\ )
But even if we fix "inner dir separators" problem we will be stopped by
expressions not supported by os2make376 and applied for make381
Przemek warned you months ago
For that reasons in some of my messages I stated my posture: I am trying
to discard os2make376 and to use only os2make381, to follow general path
of Harbour make process
We faced many situations which we are testing
>> "install" and "clean" stages show a lot of errors. At this moment we
>> are trying to get a complete build of Harbour with workarounds and
>> without problems. Then we can concentrate in "install" and "clean"
>Please be specific. What errors?
>All I'm seeing is very difficult to track reports and claims.
>For me it's extremely difficult to reassemble such puzzle from
>these bits. Maybe a complete log with a known (intact) revision
>would tell much more.
As I said, I am dealing the basic problems to build now. Later we can
review "install" and "clean" and for that reasons I do not posted that
errors and logs
Maurilio has been reviewing some of them, but as I can see, he do not
checked other cases of basic build
>I very strongly disagree with such move. We've chosen GNU
>Make to be the only one make system to build Harbour and we
>must stick to this goal.
I am following it testing with os2make381 in real OS/2 system
>You also seem to have ignored my other suggestions,
>anyhow I'd still recommend to fix "what I broke" (possibly
>intentionally!) regarding OS/2 GNU Make 3.76.
You and I know you did not intentionally :-)
Just was a consequence of a lot of changes and, as you said, "blind
changes" due unavailability of platform
This explanation is a waste of time
> * config/readme.txt
> - Deleted info on OS/2 make.
I checked what you deleted and found some interesting info
- I got my os2make381 from hobbes.nmsu.edu
- You got it from os2site
- In first deleted link are a lot of info and files of os2make
- An there are files:
MAKE-3_81-R2_LIBC063
MAKE-3_81-R2_PMW
Reviewing info we see:
- Our current os2make381 was build with gcc 2.95.3, a very old gg in OS/2
- maintainer is Andreas Buening ( andreas.buening at nexgo.de )
- Source file are in hobbes.nmsu.edu/h-search.php?key=make-3.81-r2.zip
- In February 2009 (recently) Peter Weilbacher ( mozilla at
weilbacher.org ) rebuilt the executable with GCC 4.3.3 and explained:
-------------------------
Because this now uses (and requires!) the newer C-Runtime library kLibc
0.6.3 it seems to be more stable when used in a parallel setup (e.g. -j4
as in my current Mozilla setup).
Warning: this is only a quick recompile, I did not do any serious testing!
-------------------------
Obviously I checked Harbour-Watcom without two changes of Przemek with
this new os2make381
Error was the same, at least in this case:
------------------------------------
[E:\harbour907e\harbour]make -r 1>make_ow.log
make[3]: *** [hbrtl.lib] Segmentation fault (core dumped)
make[2]: *** [descend] Error 2
make[1]: *** [rtl] Error 2
make: *** [source] Error 2
------------------------------------
Przemek:
- having the sources and possibility to rebuild os2make381, have some
benefit ?
- The problem is with os2make381, and/or OS/2 command shell (CMD.exe) ?
If later, can we use other ?
- Is any other kind of environment limit (command line size,
environment, ... ) ?
- I think if you get contact with Andreas Buening to exchange info about
these problems can help us to understand/solve them
David Macias
_______________________________________________
Harbour mailing list
[email protected]
http://lists.harbour-project.org/mailman/listinfo/harbour
