[m-dev.] Open bugs causing compiler aborts

Julien Fischer jfischer at opturion.com
Wed May 22 16:25:14 AEST 2013

Hi all,

The following bugs regarding compiler aborts are all currently still
open.  It would be good to try and get these fixed where practical. (I
am aware that in same cases the "fix" is "implement a new mode
analyser", but ...)

I have a proposed fix for #184 lying about.

#278 is just matter of working out what an appropriate error message to
emit is.  (Prior to Zoltan's rewrite of the state variable
transformation it would have been a mode error.)

I think the tabling related ones are due to bits of tabling
infrastructure being left over after dead procedure elimination has
eliminated the tabled procedure.


0000278: Compiler crash from bad state variable use.
0000272: The polymorphism pass can generate mode-incorrect code.
0000270: When one uses fact_table, the mercury compiler crashes if there is an error in the fact_table.
0000246: HLDS->MLDS code generator abort
0000184: Unresolved polymorphism in typeclass crashes mmc
0000154: compiler abort with debugging enabled
0000091: compiler throws exception in hlc.gc
0000085: stage 1 compiler aborts when compiled with -O5 --intermodule-optimization
0000086: Uncaught mode error leads to crash in codegen.m
0000071: compiler abort when building programs in debugging grades
0000060: inconsistent type_infos after type specialisation?
0000051: abort during liveness detection
0000050: another compiler abort when using debug grades
0000038: Compiler aborts in detism analysis with inst any case in negated context
0000036: another tabling bug with the high-level C backend
0000033: Compiler abort when generating low level code
0000022: compiler abort with memoing, dead proc. elimination and the hlc backend
000006: compiler abort

More information about the developers mailing list