Re: cdrecord: compiling failure on RHEL3 due to MKLINKS

[Giulio Orsero <giulioo@gmail.com>]

On Tue, 13 Jan 2009 16:36:02 +0100, Joerg.Schilling@fokus.fraunhofer.de
(Joerg Schilling) wrote:


>> ie: compiling readcd dir fails with
>>
>> make[1]: Entering directory `/build/cdrtools-2.01.01/readcd'
>> make[1]: *** No rule to make target `misc.c', needed by
>> `OBJ/i686-linux-cc/misc.d'.  Stop.

>Corect, the problem is a bug in gmake that causes gmake to ignore some
>dependencies...

>> ==== Solution
>>...
>> --- readcd/Makefile.orig    2008-12-26 21:11:15.000000000 +0100
>> +++ readcd/Makefile 2009-01-13 13:16:26.000000000 +0100
>> @@ -28,7 +28,7 @@
>>  LIBS=      -lscgcmd -lrscg -lscg $(LIB_VOLMGT) -ledc_ecc_dec -ledc_ecc
>> -lcdrdef
>> lt -ldeflt -lschily $(SCSILIB) $(LIB_SOCKET)
>>  XMK_FILE=  Mreadcd.man
>>  
>> -cd_misc.c scsi_scan.c scsi_cdr.c:
>> +misc.c cd_misc.c scsi_scan.c scsi_cdr.c:
>>     @echo " ==> MAKING SYMLINKS in ." && sh ./MKLINKS
>>  $(ALLTARGETS): cd_misc.c scsi_scan.c scsi_cdr.c
>
>To you like me to tell that adding other dependencies will work around the 
>gmake problem?

I don't understand the question.

I'm just saying that by explicitly writing all the symlinks names in the
Makefile MKLINKS rule the gmake problem disappears.

I'm not an expert in makefiles to say if this is an acceptable solution that
you may want to use if you want compatibility with gmake or if it has
negative side-effects or is in any case superfluous for "non gnu" make so
that you won't add it.

Since I have a solution for this specific problem, whatever you choose is
fine by me.
-- 
giulioo@pobox.com