stdlib won't compile
Bugzilla Link | 1153 |
Created on | Apr 18, 2015 15:08 |
Version | svn |
OS | Linux |
Architecture | PC |
Extended Description
I encounter the same problem. I use repeated "make -j8" invocations get the rest of the stdlib to compile to the point where it is usable by sac2c. The damage happens very early, as can be seen when I tried to fault-isolate the problem: sac2c UnibenchCSV.sac -v1 -bptc:rso >crud Internal compiler error Assertion "TCcountExprs( exprs) == 1" failed at print/print.c:1136 -- operator application with wrong number of arguments encountered! This is an area of the compiler that I know nothing about, so my investigation stopped here. Perhaps Artem will look into it, as it might be a parser problem. Bob On 15-04-18 07:01 AM, Heinz Wiesinger wrote: > Hi all, > > I tried to update to sac2c master today, and found that it is unfortunately > again broken, in the sense that it can't compile stdlib. This is the error I'm > getting: > > cd modules/unibench/; sac2c -v0 -linksetsize 0 -O3 UnibenchInput.sac -o > /mnt/progs/projects/UvA/sac//stdlib/shared-libs > Internal compiler error > Program reached impossible state at arrayopt/lacfun_utilities.c:466 -- Could > not find relational for predicate > Please file a bug at: http://bugs.sac-home.org > make: *** [libUnibenchInputTree.so] Error 1 > > I can finish compiling as much as possible with "make -k", but eventually I hit > the same bug in my test application. I would appreciate if someone could take > a look at this because I have no idea what's going on here. > > As a sidenote, is it somehow possible to get continuous integration back for > sac2c/stdlib? I'm developing on top of a 2 month old sac2c simply because > everytime I try master it's broken in one way or another. > > Grs, > Heinz This has been broken for some time now, but it definitely breaks with: sac2c -V sac2c v1.00-beta (Haggis And Apple) developer rev 18553 linux-gnu_x86_64 (Fri Apr 17 16:28:28 EDT 2015 by sac)