[Zlib-devel] zlib 1.2.4.3 for your perusal

[Enrico Weigelt]

* Volker Grabsch <vog at notjusthosting.com> schrieb:

> How can a completely optional feature break anything?
> ... in a build environment as simple as that of zlib?

It changes semantics, as soon as somebody's set this variable.
(yet another one which has to be filtered out).
 
> The $CROSS_PREFIX's only task is to initialize $CC, $LD, $AR, etc.
> How complicated can that be, and what kind of builds does this
> break? Any system on which $CROSS_PREFIX causes wrong initializations
> can set the $CC, etc. by hand.

The whole idea is unclean. Build systems generally should pass
all toolchain commands explicitly, which already is the de-facto
standard. These target-type prefixes are only work for the specific
case where you have only one toolchain per target-type.

Just look at crosstool for a sane way ;-p

> I fully agree. In mingw-cross-env, my experience is that most
> packages conflate the build with the host environment, mostly
> visible by _running_ test scripts (instead of just building them)
> and by calling tools like "uname" to get host environment information.

ACK. That's a general misdesign.
Sometimes it get's even worse: building programs which generate
source code or data files on the host system (of course built w/
$CC instead of $HOST_CC).


cu
-- 
---------------------------------------------------------------------
 Enrico Weigelt    ==   metux IT service - http://www.metux.de/
---------------------------------------------------------------------
 Please visit the OpenSource QM Taskforce:
 	http://wiki.metux.de/public/OpenSource_QM_Taskforce
 Patches / Fixes for a lot dozens of packages in dozens of versions:
	http://patches.metux.de/
---------------------------------------------------------------------