[Ecls-list] Build cleanups, part one.

Juan Jose Garcia Ripoll jlr at mpq.mpg.de
Tue Mar 29 01:25:16 UTC 2005

Hi Maciek,

first of all, thanks for your contribution. I had a look at your patches 
and tried them. There are some serious problems with them. First of all, 
you have merged the LIBS and CLIBS variables. This breaks the 
configuration process, because LIBS contains the garbage collector and, 
since the garbage collector is not yet built, some test programs will 
fail at configuration time.

The reason is that LIBS and CLIBS are not the same thing. CLIBS are C 
libraries which are required for building both the bootstrapping code, 
ecl_min, the core library, libecl.so, and also all modules. This 
variable may contain libraries which are built at compilation time and 
hence it cannot be merged with LIBS.

STATICLIBS has a similar value, but only for statically linked libraries 
which are linked to form libecl.so or ecl_min and are not needed by the 
modules (*.fas files).

Could you please fix your patches to take this into account: either 
restoring the uses of CLIBS and STATICLIBS or introducing three new 
* LIBS for all C libraries that are installed and are required by all 
code (like -ltcp, -lsockets...)
* FASL_LIBS for C libraries that are required by all ECL code (like 
-lgmp, -lgc, etc)
* CORE_LIBS for C libraries that are required only by ecl_min and 
libecl.so (like -leclgmp, -leclgc,...)

The old CLIBS would correspond to LIBS + FASL_LIBS, while the old 
STATICLIBS would correspond to CORE_LIBS.

Also, I am not sure about the verbose --with-system-boehm-includes, etc, 
etc. There is a canonical way to specify include and library 
directories, which is to prepend your call of configure with a new value 

CFLAGS="-I/usr/local/include" LDFLAGS="-L/usr/local/lib" ./configure 

Did the FSF guidelines change to this respect?

Regards, Juanjo

-------------- next part --------------
A non-text attachment was scrubbed...
Name: jlr.vcf
Type: text/x-vcard
Size: 361 bytes
Desc: not available
URL: <https://mailman.common-lisp.net/pipermail/ecl-devel/attachments/20050329/23beeb9e/attachment.vcf>

More information about the ecl-devel mailing list