greg@mobius.Viewlogic.COM (Gregory Larkin) (11/16/90)
Hi all,
I am busily trying to upgrade to V1.5.10 from V1.3 on my PC. From
what I had heard from the net, I figured it might be a good idea to
install estdio as part of the upgrade. I have patched /usr/include
to V1.5.10, so estdio and libc.a are the next logical step.
I ran the yinstall.sh script successfully. Should I have used
yminix.sh instead, if a am using a standard V1.3 minix with the ACK
compiler??
I did a "make stdio" after the script. All of the sources were
compiled correctly, but tsort failed with the message "Error: loop
in the input", or something like that. I *do* have a non-zero
length stdio.a, though, so something happened. Is this really an
error?? If it is, what is the way to fix it?
PS: I also had to chmem +30000 for lorder and tsort.
Thanks for the help,
--
Greg Larkin (ASIC Engineer)
Viewlogic Systems, Inc. (The CAE Company)
293 Boston Post Road West -------------------------------------
Marlboro, MA 01752 |"We've got captains not courageous,|
508 480 0881 x321 |captains tumbling into madness..." |
E-mail: greg@Viewlogic.COM |Peter Garrett, Midnight Oil |
-------------------------------------cechew@bruce.cs.monash.OZ.AU (Earl Chew) (11/19/90)
In <1990Nov16.093147@mobius.Viewlogic.COM> greg@mobius.Viewlogic.COM (Gregory Larkin) writes: >I ran the yinstall.sh script successfully. Should I have used >yminix.sh instead, if a am using a standard V1.3 minix with the ACK >compiler?? No. I thought that I had made this clear in the documentation. Seems not to be the case since you're the second person within the last few days to be stumped at this point. 1. Everyone should firstly run yinstall.sh and select the ack site description (if they're using ack) or the bcc site description (if they're running bcc). 2. Everyone should then make stdio to ensure that everything compiles without fuss. 3. Everyone should then make test to ensure that all the test programs can be compiled without fuss. 4. Everyone should then run TEST/exercise to ensure that estdio has been compiled ok. 5. *NOW* comes the point where those with Evans 386 should edit makelib as described and use sh makelib [86|386], while those with standard Minix ack can try to use yminix.sh (or roll their own). >I did a "make stdio" after the script. All of the sources were >compiled correctly, but tsort failed with the message "Error: loop >in the input", or something like that. I *do* have a non-zero >length stdio.a, though, so something happened. Is this really an >error?? If it is, what is the way to fix it? Tsort is probably the most broken program in 1.5.10 :-( There should be no loops (that I can remember). Try 1.5.9 tsort, or perhaps the new `better' tsort that was posted to the net. I use 1.5.9 tsort, and haven't tried the new tsort. If the copy of tsort you're using is ok, and you're still getting an error message, then something else is wrong. Try more memory for tsort. >PS: I also had to chmem +30000 for lorder and tsort. lorder also needs work. It uses a stupid strategy for processing files where a separate libupack process is forked for each separate input file. This takes *lots* of time (even with a 386). I hacked lorder to fork libupack *once* and feed it all the .s files concatenated together (ie fake a .a file). This makes running times of lorder more acceptable. Earl -- Earl Chew, Dept of Computer Science, Monash University, Australia 3168 EMAIL: cechew@bruce.cs.monash.edu.au PHONE: 03 5655447 FAX: 03 5655146 ----------------------------------------------------------------------