On Wed, 4 Apr 2001, Forest Dean Feighner wrote:
> Peter,
>
> Thanks for the reply.
>
> When I ran ./configure and make from within the ImageMagick directory
> the make process tried to make PerlMagick as well. The README, in regard
> to MAP_TARGET, says:
>
> To create and install a new 'perl' executable (replacing your
> existing PERL interpreter!) with PerlMagick statically linked
> (but other libraries linked statically or dynamically according
> to system linker default), execute
>
> perl Makefile.PL
> make perl
> make -f Makefile.aperl inst_perl
>
> or to create and install a new PERL interpreter with a
> different name than 'perl' (e.g. 'PerlMagick') and with
> PerlMagick statically linked
>
> perl Makefile.PL MAP_TARGET=PerlMagick
> make PerlMagick
> make -f Makefile.aperl inst_perl
>
> This worries me a bit because I get the impression that it'll replace
> the perl exectutable with PerlMagick.
I do not think that it will replace perl with PerlMagick it will leave
perl in your default bin directory then add the PerlMagick next to it
in the same directory. If you had specified:
perl Makefile.PL MAP_TARGET=perl
make
make -f Makefile.aperl inst_perl
then that incantation would have made a new binary named 'perl' with the
Image Magick library statically linked in.
The question then becomes "why bother with static linking at all?".
Perhaps in the case of a graphics intensive thing like Image Magick there
could be some sort of speed advantage. But OS X supports dynamic loading
and you perl should too. What do the following return on your system (and
are you using the system supplied perl or did you build it yourself?):
perl -V:usedl
perl -V:so
perl -V:dlsrc
> After failing the make process I ran make clean, and added
> -L/System/Library/Perl/darwin/CORE/ to the LIB line in Makefile.PL.
>
> In accordance to the README;
> To create and install the dymamically-loaded version of
> PerlMagick (the preferred way), execute
>
> perl Makefile.PL
> make
> make install
>
>
> I am still having difficulties with graphic libraries, but for now, I'm
> going to take it one hurdle at a time.
Ah - OK. That looks better to me (i.e. use dynamic loading). Thanks for
your detailed reply.
Peter Prymmer
