From: "Dr. Stephen Henson" <st...@openssl.org>
> [...]
> > What is the justification for having mixed-/upper-case options/values
> > like, say:
> >
> > -CAfile
> > -CApath
> > -EncryptedData_encrypt
> > SMIME|PEM|DER
> > ???
>
> It has never been mentioned as being a problem before. [...]
No one (practically) notices _build_ failures on VMS, so it's not
amazing that a bunch of case-related (mere) _test_ failures went
unnoticed for some time. More important, tests on VMS are generally run
from VMS-specific (".com") command procedures, and adding quotation to
those is easy enough. Adding all the annoying command-line-argument
quotation to a ("portable") perl script would annoy a much wider
audience.
Note that the original complaint in the Subject ('[...] using
"extended parse-style"') was partly accurate. Jouk and I both operate
with SET PROCESS /PARSE_STYLE = EXTENDED, but he also had a CRTL
"feature logical name" defined, DECC$ARGV_PARSE_STYLE, and that was why
he saw the build problem and I didn't. With DECC$ARGV_PARSE_STYLE
defined, the process parse-style setting controls case preservation of a
C-program command line; without it, the parse-style setting has no
effect on case preservation. Everything's complicated, and adding case
preservation/sensitivity to an OS which traditionally lacked it is more
complicated than some other things. (And the original (DEC) decision to
have the CRTL down-case everything (because UNIX code always uses lower
case) might now be seen as shortsighted.)
> [...] There have been case
> sensitive command line utilities in OpenSSL/SSLeay from the beginning. The
> -CAfile and -CApath names for example are common across all the appropriate
> utilities.
It was (I claim) not a clever decision, irregardful of when it was
made. Because of the VMS CRTL down-casing, case _sensitivity_ is not a
problem, so long as everything's all-lower-case. Mixed-/upper-case
command-line stuff tends to need extra help (and on VAX on older non-VAX
systems, and, apparently, in a perl system() function, that help may be
limited to annoying extra quotation).
> That's no reason to retain them of course and just changing to a case
> insensitive comparison algorithm (strcasecmp) in the relevant places (or just
> affected platforms) would address that.
I wouldn't object. Actual users might not care, of course.
Note, too, that adding that recently suggested apps/vms_decc_init.c
module effectively sets DECC$ARGV_PARSE_STYLE inside a program, which
makes this stuff slightly less painful (on the newer non-VAX systems
where it's effective, at least). (It must be set very early in the
image activation sequence, hence the LIB$INITIALIZE complexity.) But I
wouldn't rip all those new quotation marks out of the test procedures.
("Remember the VAX!", I always say.)
------------------------------------------------------------------------
Steven M. Schweda s...@antinode-info
382 South Warwick Street (+1) 651-699-9818
Saint Paul MN 55105-2547
______________________________________________________________________
OpenSSL Project http://www.openssl.org
Development Mailing List openssl-dev@openssl.org
Automated List Manager majord...@openssl.org
