Xref: utzoo news.software.b:3641 comp.sources.d:4410 Path: utzoo!utstat!jarvis.csri.toronto.edu!mailrus!uwm.edu!rpi!tale From: tale@pawl.rpi.edu (David C Lawrence) Newsgroups: news.software.b,comp.sources.d Subject: Re: C News "upact" in Perl 3.0 Message-ID: <256C5FD7.6CCB@rpi.edu> Date: 23 Nov 89 21:23:35 GMT References: <2565B57C.10978@ateng.com> <1989Nov19.014731.26346@utzoo.uucp> <25682EB0.24425@ateng.com> <256AF479.21036@ateng.com> Organization: Rensselaer Polytechnic Institute, Troy NY Lines: 29 In <256AF479.21036@ateng.com> chip@ateng.com (Chip Salzenberg) writes: Chip> I suspect that the problem is that your Perl doesn't know how to Chip> readdir() correctly. Before blaming the script, it might be Chip> good to double-check your Perl with a small test program. I'm sorry you felt like my posting was an affront on your programming ability, Chip, but I am quite certain that any such idea was inferred and not implied. If you had read what I said, $low was being correctly set in the loop. This couldn't happen if my readdir() is broken, which it is not. I can blame the script, because if I hadn't run the script it wouldn't have munged my file. The fact that the script is coded fine, which is why I _did_ spend time tracking down what was happening, does not change the damage done. Larry suspects the problem is caused by the GCC optimizer. In mail, J Greely had this to say, cited with permission: J> Patchlevel 6 tickled to life an old bug with return and local J> variables. I'm trying to clear it up with Larry, but my examples are J> twisted (you have to have a local, a while, *and* a return, and the J> body of the while can't be empty, and *must* refer to a local, and J> when the moon is full...). Once again, I am sorry that you felt I was making a slur on your programming ability. I was not. Dave -- (setq mail '("tale@pawl.rpi.edu" "tale@ai.mit.edu" "tale@rpitsmts.bitnet"))