Sender: nate AT cartsys DOT com Message-ID: <3709227A.26655F3A@cartsys.com> Date: Mon, 05 Apr 1999 13:52:10 -0700 From: Nate Eldredge X-Mailer: Mozilla 4.08 [en] (X11; I; Linux 2.2.5 i586) MIME-Version: 1.0 To: "Salvador Eduardo Tropea (SET)" CC: djgpp-workers AT delorie DOT com Subject: Re: YAMD tests References: Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Reply-To: djgpp-workers AT delorie DOT com Salvador Eduardo Tropea (SET) wrote: > > Hi all, specially Nate: > > I tried YAMD and here are some comments, I'm sending to the list because > perhaps others knows why some things are behaving strange. > > 1) There are a bug in the Makefile (I think Alex will report it too, we both > tested the package here at INTI). Nate: you used DJGPP as variable in make > and it overwrites the DJGPP environment variable, use something like isDJGPP > instead. > Question: Why Nate didn't see it? is because he used bash? I'm wondering that myself. No, I didn't use bash. The environment variable is clearly affected, but GCC ran fine nevertheless. Oh well, it's now fixed. > 2) test9 (touching a byte out-of bounds in a constructor fails). We tried gcc > 2.7.x and 2.8.0. I found that too. It's because the initialization is done in a constructor and I can't ensure that it's the first constructor to run. I tried forcing my code to be linked first; that makes my destructor first, not last. Can't win :) So what I need is some way to get my code run first and last. Any ideas? > 3) test10 enters in an infinite loop, the traceback is endless. Now *that* I haven't seen. I did change a couple of other things for 0.22 (just uploaded) that could theoretically be related, but I'm pretty sure in real life they aren't. Is it my traceback or the standard DJGPP one that loops? Maybe you could grab 0.22 and do it again, and then send me the output (symified with `symify' or `do-syms' as appropriate, if possible). Thanks. -- Nate Eldredge nate AT cartsys DOT com