Subject: Re: CVS commit: src/crypto/dist/openssl
To: Perry E. Metzger <email@example.com>
From: None <firstname.lastname@example.org>
Date: 07/15/2003 18:24:54
>> > Log Message:
>> > unifdef VMS/WIN16/WIN32 in public headers, at least
>> >is this actually necessary? "dist" sources should not be changed
>> >to be less portable for no reason... the only real change i can
>> >see this makes is making it harder for future openssl updates due
>> >to unnecessary conflicts....
>> header files shouldn't have #ifdef which easily conflict with what user
>> would define by chance. there's no portability issue, we won't and
>> shouldn't use the removed prototypes (and whatever) anyways.
>I agree with Matt that the dist headers should remain the
>same. However, you are right that /usr/include header files should be
>clean. Perhaps unifdef could be called on the untouched files at build
>time, so that clean files are installed even though the originals
i fear possible inconsistencies while building libcrypto (with all
#ifdef VMS intact) and installing include files (with all #ifdef VMS
removed) with the above approach.
another problem is that unifdef(1) does not support "#if defined(foo)"
form. it isunderstandable as we can put very complex expression
using this form, and partially evaluating them would be difficult