blob: 5a75e5b57ac1ce4304ff0f5849a8a3abcc39b3ac [file] [log] [blame]
(Message inbox:40)
Date: Tue, 17 Jul 1990 11:56 EDT
From: swl26%CAS.BITNET@CORNELLC.cit.cornell.edu
Subject: Re(2): port of flex-2.3 to IBM/MVS
To: vern@cs.cornell.edu
>Message received. I'm sending this response to
>swl26%CAS.BITNET@CORNELLC.cit.cornell.edu.
Great -- we can talk.
>How extensive are the diffs? If they're fairly short then yes, please
>send them. If they're pretty involved then probably not, as I'm not aware
>of many MVS flex users ...
I've built a context diff file which is ~850 lines.
Summary of major changes necessary:
o a new initscan.c -- because MVS is ebcdic, the initial scanner had
to be changed. I built a scanner from scan.l using -Ce, then
hand-manipulated the yy_ec table to fit EBCDIC instead of ASCII
chars. (This is not included in the diff file above.)
o changes in main and flexdef.h to change how files are handled. (No
dynamic file names, etc.)
o Some lines had to be shortened to 80 bytes. This mostly impacted
your RCSID lines, which with the full path name were too long.
o SASC and the linker don't allow externals to be longer than 8 chars.
I thus wrote a Q&D program to shorten all externals. (This would be
a separate file 'fixit.l', which is not included in the diffs.)
o There are several places where there are tests for 'isascii' (which I
deleted conditionally, using #ifdefs), and tests for 'A' <= x <= 'Z'
which I changed to 'isupper'.
Many of the changes I've incorporated without impacting other systems.
Others are with 'ifdefs'. Still others (the short external names and new
initscan) are 'isolable' as separate files which would have to be
included with a distribution.
Obviously, you are not going to want to (even if you can :-) ) worry about
testing new versions in the MVS environment. Likewise, I don't have the
time or resources to provide much support. (I'm not sure my management
would allow any :-( )
With all of the above feel free to say "Thanks, but no thanks." If you
want to see the diffs anyway, I will certainly mail them.
Thanks for your time and efforts.
Steve Layten
Chemical Abstracts Service, PO Box 3012, Columbus, OH 43210, +1 614 421-3600
INET: swl26%cas.BITNET@CUNYVM.CUNY.Edu