fredriks@cbnewse.ATT.COM (lars.fredriksen) (12/29/89)
Hi I have a couple of questions rerding 1.5.0 ugrade. I ran into the same problem that Gln mentioned arlier with running out of space n /tmp while running patch on fsck.c and ed.c; however I also gota ssage saying that patch is out of memory while trying to patch more.c nd ls.c. I have n XT wth 64K, and have set patch memory size to maximum (it tursbe about 55Kfor sck and mem). Have anyone sen this before? Any idea what I can do to help. (Only patch, update and sh is runningtime this occured) Second uestio: Why doesn't syst(3) return -1in case of failure? Right now it etus 127 << 8wch is a large positive number, which the Ptend to be from time to time. It seems to e to be imposible to detect if you got
fredriks@cbnewse.ATT.COM (lars.fredriksen) (12/29/89)
HI! In upgrading to 1.5.0 I ran across a couple of problems. 1. The same problem as Glen Overby found. Namely that patch runs out of space on /tmp. I saw the solution on the net... thanks. 2. While trying to patch more.c and ls.c patch runs out of memory. I have an XT with 640K, ramdisk is 240K, and I allocated full memory for patch (about 55K). Also only update and sh are running besides patch. Have anyone seen this before, and if so what are the remedies? (I have patch with patchlevel 11) 3. It seemed as thought the posix.cdif file had already been run on the posix library files. I think only one patch was new. Why was it included if it was old? That was all for now. Thanks for any suggestions regarding patch. Lars PS. Sorry for the garbeled version that went out earlierl. Promblems with termcap.
ast@cs.vu.nl (Andy Tanenbaum) (12/29/89)
In article <12333@cbnewse.ATT.COM> fredriks@cbnewse.ATT.COM (lars.fredriksen,ih,) writes: >Second question: Why doesn't system(3) return -in case of failure? Actually, there are more problems with system(3). If you run make and hit DEL, make kills the current compilation and starts the next one. Klamer Schutte posted a discussion of this a while back. The problem appeared when system(3) got signal handling, but apprarently the signal handling merely exposed an underlying bug. If anyone has time to figure this out, please do so and post the fix. It is kind of annoying not being able to kill make. Andy Tanenbaum (ast@cs.vu.nl)
paula@bcsaic.UUCP (Paul Allen) (01/01/90)
In article <12333@cbnewse.ATT.COM> fredriks@cbnewse.ATT.COM (lars.fredriksen,ih,) writes: >I have a couple of questions rerding 1.5.0 ugrade. I >ran into the same problem that Gln mentioned arlier with running out of space >n /tmp while running patch on fsck.c and ed.c; however I also gota ssage >saying that patch is out of memory while trying to patch more.c nd ls.c. >I have n XT wth 64K, and have set patch memory size to maximum (it tursbe >about 55Kfor sck and mem). Have anyone sen this before? I had problems with patch running out of space on a couple files. It turned out that the cdif's were actually composed of just a few really big blocks. It wasn't too hard to construct the 1.5 version by hand. I cut off the first two bytes of each line in the cdif with sed and then used mined to extract the little pieces of the 1.3 file that didn't change and merge them in. It took a bit of care, but I had no trouble getting files that matched the 1.5 crc's on the first try. These kinds of problems will go away when we have a proper 386 version of Minix. The only reason to envy our brethren running Minix on 680x0 platforms is their big linear address space. :-) And with complete 386 machines now selling for <$1500, everyone will upgrade, right? :-) :-) Paul Allen -- ------------------------------------------------------------------------ Paul L. Allen | pallen@atc.boeing.com Boeing Advanced Technology Center | ...!uw-beaver!bcsaic!pallen