Recent

Author Topic: GitHub runner macos-latest and Lazarus 2.0.10 not working  (Read 3696 times)

Gustavo 'Gus' Carreno

  • Hero Member
  • *****
  • Posts: 1114
  • Professional amateur ;-P
GitHub runner macos-latest and Lazarus 2.0.10 not working
« on: May 23, 2021, 11:44:51 pm »
Hey all,

I'm really not sure where or to whom I should report this:
https://github.com/gcarreno/laz-JSON-Viewer/runs/2597532516?check_suite_focus=true

What baffles me is the fact that GitHub runner macos-latest and Lazarus 2.0.12, have no issues:
https://github.com/gcarreno/laz-JSON-Viewer/runs/2597532492?check_suite_focus=true

So should I create a Bug report?

Thanks in advance!!

Cheers,
Gus
Lazarus 3.99(main) FPC 3.3.1(main) Ubuntu 23.10 64b Dark Theme
Lazarus 3.0.0(stable) FPC 3.2.2(stable) Ubuntu 23.10 64b Dark Theme
http://github.com/gcarreno

Martin_fr

  • Administrator
  • Hero Member
  • *
  • Posts: 9794
  • Debugger - SynEdit - and more
    • wiki
Re: GitHub runner macos-latest and Lazarus 2.0.10 not working
« Reply #1 on: May 24, 2021, 02:13:59 am »
The problem is from the log I can not tell which app is crashing
Code: [Select]
(9001) Assembling (pipe) /Users/runner/.lazarus/lib/SynEdit/units/x86_64-darwin/cocoa/allsynedit.s
(1008) 111423 lines compiled, 18.9 sec
(1021) 18 warning(s) issued
(1022) 87 hint(s) issued
An unhandled exception occurred at $00000001032A7CD2:
EAccessViolation: Access violation
  $00000001032A7CD2
  $00000001037594D0
  $000000010375D4D4
  $000000010375E6EE
  $00000001031AB043
First guess would have been the compiler.
But "hint issued" compilation just finished.

If it were in the middle of code, it could be a compiler bug. And even though it is the same 3.2.0 compiler as in the successful run, it does compile a different source, as units have changed.
But after compile finished, it is extremely unlikely the compiler would crash depending on the units content.

So that would make "lazbuild" the more likely candidate.

It is possible that there were changes to lazbuild (or units used by it) between the versions. It is also possible, that the bug (if indeed in lazbuild) was there, and got triggered by changes in the output of fpc, or the directory structure....
I do not know, and it is also possible that lazbuild (if indeed..) is fixed upstream....


Unfortunately, afaik FPC does not support -gl (dumping of line numbers and unit names) on Mac.
So not sure how to get more / better info.

If lazbuild can be build with debug info, and the script could some how run it in gdb, then that may be a way to get a stacktrace....
(and a stacktrace may be able to confirm if it is lazbuild)

Better to try that on a local PC.

 

TinyPortal © 2005-2018