home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
Cream of the Crop 20
/
Cream of the Crop 20 (Terry Blount) (1996).iso
/
os2
/
uupcdiff.zip
/
uupc_fix.doc
< prev
next >
Wrap
Text File
|
1996-07-14
|
4KB
|
66 lines
I have uploaded to ftp-os2.nmsu.edu and some of its mirror sites
UUPC_FIX.ZIP and UUPC_DIFFS.ZIP. These files contain a fix for the
stdin of inews.exe, which reduces compatibility between UUPC/Extended,
Changi and Souper.
Many months ago, I posted an email to the UUPC list detailing that I
had doubts about the Changi authors claim that standard input did not
work correctly in UUPC. Even though I doubted his claim, I had no proof
either way, and I was reluctant to delve into the matter. I don't use
IBM's C-Set, and I already have more compilers than a sane person can
take. However, neither Watcom nor emx, the only two left now, have
shown that they care much for compiling UUPC code. After souper came
out with another update, however, I was forced to rethink how I wanted
to attack the problem. The answer was obvious: standard input does
work in inews and rmail, but it obviously isn't receiving the data that
it expects. I can either try to keep up with the authors of changi and
souper, or meet the challenge head on.
Therefore, for about the fourth time, I slugged my way through the UUPC
code and got it to compile. Then, after dumping core several times,
I got it running respectably for the first time ever. I then found
out the version info, etc., cleaned it up and started looking for the
real problem. It jumped out at me after only an hour or so. I tried
it out on the original changi and souper executables. Changi worked,
but Souper did not. I recompiled Souper, because the error did not
make sense, and Souper worked as well.
Therefore, I have come to the following conclusions:
1. stdin has always worked in inews and rmail.
2. The author of souper made change in his code, which caused rmail
to work correctly with souper without any code changes. However,
since he changed compilers, a recompile is still necessary or it
will look for the POP3 server instead of MAILER environmental
variable. (I have looked at reply.c to verify this fact)
3. While stdin has always worked in inews, the code itself caused
it to appear to not do so, as it was searching for headers in a par-
ticular order. The stdin file was OK, in other words, but the
underlying code was trying to force an issue that it should not
have. The same exact code on a temporary file works just fine.
4. There are way too many #define statements in UUPC to make any
of this an easy task.
If you wish to verify these changes yourself, but don't want the
entire package, then simply download UUPC_DIFFS.ZIP. If you want
the entire fix (including the diffs), download UUPC_FIX.ZIP.
Executables are included in the latter. This is not a replacement
for the UUPC/Extended package, and it only contains enough files
to recompile inews and rmail and for someone else to verify that my
observations are correct.
I will upload these tonight to Hobbes. I will attempt to upload
them to ftp.cdrom.com and ftp.leo.org as well, but the first is
usually busy and the second has a very slow link (for me). I will
only post the diff for inews to the UUPC/Extended mailing list,
but I will provide the entire fix to Kew Software upon request.
While I occasionally do programming for my job, my profession is
not a programmer. Most of it is in my spare time (even when it
is for work), so criticisms of my style of programming will be
sent to dev/null. However, inaccuracies in my above observations
will be weighed according to the evidence presented to me.
John D. Carmack