Date: Sat, 28 Jul 2001 10:21:27 +0300 From: "Eli Zaretskii" Sender: halo1 AT zahav DOT net DOT il To: "Andrew Cottrell" Message-Id: <7263-Sat28Jul2001102126+0300-eliz@is.elta.co.il> X-Mailer: Emacs 20.6 (via feedmail 8.3.emacs20_6 I) and Blat ver 1.8.9 CC: djgpp-workers AT delorie DOT com, sandmann AT clio DOT rice DOT edu In-reply-to: <005b01c116fc$6d4da7e0$0a02a8c0@acceleron> (acottrel@ihug.com.au) Subject: Re: Win2K testing, WinXP testing References: <10107262018 DOT AA16742 AT clio DOT rice DOT edu> <005b01c116fc$6d4da7e0$0a02a8c0 AT acceleron> Reply-To: djgpp-workers AT delorie DOT com Errors-To: nobody AT delorie DOT com X-Mailing-List: djgpp-workers AT delorie DOT com X-Unsubscribes-To: listserv AT delorie DOT com Precedence: bulk > From: "Andrew Cottrell" > Date: Sat, 28 Jul 2001 10:29:55 +1000 > > Are the DJGPP packages stock standard from Simtel or have you re-built > them? Stock versions, for now. I don't think we've got to the place where changes in the current CVS should matter. > The two unit test I can think of that are still on my to do list for Windows > 2000 are: > 1) Get make working > 2) Get rhide working > Once these two unit tests are done then the other testing that needs to be > done is: > 1) See if our own programs that we have written work on 2000 or XP. > 2) Re-build all of the other standard GNU packages > 3) Use all of the re-built GNU packages to then build some of the V2TK, > V2APP packages and see if they work. Based on previous experience, I'd suggest to also test these aspects: - debugging - profiling - Ctrl-C and Ctrl-BREAK (do they do what you expect or crash?) - programs which catch SIGFPE and SIGINT Basically, go through all the gotchas described in section 3.3 of the FAQ and see what happens on W2K and XP. Btw, someone reported a few months ago that Emacs abruptly exits on NT if you press and hold an arrow key right after typing "emacs RET" from the command line (see the thread "Re: Emacs crashes with fast arrow keys"). It would be nice if someone could see if the same happens on W2K and XP. > 1) Make 3.791: > a) shells out to using command.com via spawnvpe function > in job.c approx line 1200 (+/- 30 lines due to debugging ) to execute > "command.com /C if not exist glob\nul md glob" then the command.com displays > an errror 183, but if I use cmd.exe then I do not get the error. > I need to check at work on NT 4.0 if the same > problem exists and if it does then see if I can produce a patch for job.c so > under Windows NT make uses cmd.exe instead of command.com. I think this is not the right fix (I can't be sure it is possible to invoke cmd.exe on any given system). I think we should simply redirect the error message to /dev/nul, or maybe find another way of testing if a directory exists.