On Sat, 20 Apr 2002, Richard Levitte - VMS Whacker wrote:
> I've looked at the two patches you sent (relative to 0.9.6d and
> 0.9.7). Unfortunately, I believe that the 0.9.6d won't get applied,
> because of unpredictable problems that might occur. Therefore, I will
> only comment that 0.9.7 one.
I see the point of not applying to 0.9.6d right before a release. At
least the patch is in the archives for anyone who wants to use it.
> I understand the reasons behind several things you do, like the
> symlink hackery. However, I can't help but wonder why you do that at
> all in Configure when util/mklink.pl already deals with the situation,
> and would therefore properly take care of it when 'make links' is run.
The problem is that util/mklink.pl only works with algorithms that
are included in the build. It is called from the makefile in each
subdirectory. When algorithms are excluded (e.g., by configureing with
no-idea), then mklink.pl is never called for the excluded directories
(e.g., crypto/idea). The symlink alternatives in Configure are made
only for the excluded algorithms (@skip). The other changes are in
mklink.pl. If you don't make the links for the excluded algorithms,
there will be errors, at least with the doing "make test". You may not
see this if your tar program makes the symbolic links on unpacking the
archive.
> Does system() not work in perl under DOS? It seems like you want to
> avoid using system() as much as you can. If not, then why the hackery
> of util/mklinks.pl?
System does work under DJGPP. Depending on how it is called, it may
call bash or the DOS command.com file. system() should work (I think
I submitted the original patch that put it there), but this assumes
that compatible utilities are on the system. It seemed more robust to
keep all the work within perl, rather than assuming that the correct
utilities are there.
> Now, in certain test scripts, you have added lines that very much
> remind me of C preprocessor lines. Care to explain that, and
> especially how you see them getting processed as you intended? And
> just so we're clear on this, I am not willing to patch sh scripts with
> some cpp hackery like that. Been there, done that, don't wanna do it
> again.
Good point. I hadn't changed this since the first patch I submitted in
February 2001. I think this was simply an error on my part. I am not
sure why it seemed to work. I will try to rework this and resubmit,
so that the path manipulation works in the shell scripts. Thanks for
pointing out this basic error.
Doug
__
Doug Kaufman
Internet: [EMAIL PROTECTED]
______________________________________________________________________
OpenSSL Project http://www.openssl.org
Development Mailing List [EMAIL PROTECTED]
Automated List Manager [EMAIL PROTECTED]
