"error: Success" from fsm_trim

[katef]

; cat /tmp/ha-target.fsm 
0; 1; 2; 3; 4; 5;

0  ->  1;
0  ->  1 "e";
1  ->  2 "b";
2  ->  3 "g";
2  ->  4 "h";
3  ->  0 "h";
4  ->  0;
4  ->  1 "h";
5  ->  1 "a";
5  ->  3 "h";

start: 5;
end:   3;

gives:

; fsm -t trim -capl dot < /tmp/ha-target.fsm
fsm_op: Success

[katef]

and:

; cat /tmp/ha-target.fsm 
0; 1; 2; 3; 4;

0  ->  2 "a"; # e.g. "a"
0  ->  1 "b"; # e.g. "ac"
0  ->  0 "c"; # e.g. "ac"
1  ->  0 "c"; # e.g. "ac"
1  ->  2 "c"; # e.g. "a"
2  ->  0 "c"; # e.g. "ac"
2  ->  1 "c"; # e.g. "ac"
3  ->  2 "a"; # e.g. "a"
4  ->  3 "a"; # start
4  ->  2 "c"; # e.g. "a"

start: 3;
end:   1;
; fsm -t trim < /tmp/ha-target.fsm 
zsh: segmentation fault  fsm -t trim < /tmp/ha-target.fsm

[katef]

smaller:

; cat /tmp/ha-target.fsm 
0; 1; 2; 3; 4;

0  ->  1 "g";
1  ->  2 "b";
2  ->  1;
3  ->  3 "h";
4  ->  0 "b";

start: 4;
end:   0;
; cat /tmp/ha-target.fsm | fsm -t trim
zsh: done                cat /tmp/ha-target.fsm | 
zsh: segmentation fault  fsm -t trim
;                                     

[silentbicycle]

Is this about the return type from fsm_trim? I have an instance or two of that in test code fixed on my branch for captures, but if I found any in the main code I thought I'd posted a PR to fix them already. I can look into this next week.

At whatever point the entire codebase switches to C99, checking whether return types need to be int rather than bool may find issues.

[katef]

I'm not sure about the "error: Success", but the segfault is around line 324:

        base = (s_id == 0 ? 0 : offsets[s_id - 1]);

where for the most recent test case there:
offset_count = 3
s_id = 4

[silentbicycle]

I'm pretty sure I have that specific issue fixed on my branch, I'll cherry-pick it out and post a PR in the next couple days. I thought I had introduced it as part of work on that branch, but apparently not.

[katef]

Confirmed that #356 fixes both issues here.