Feature request @ "too few cycles" after filtering

[Andi47]

I just found this From an aliquot-sequence run using aliquit (which used factMsieve.pl):

[Dec 27 2009, 03:10:43] c123: running gnfs (ggnfs)...

<after sieving, doing a filtering run:>
...
Mon Dec 28 18:47:09 2009  commencing 2-way merge
Mon Dec 28 18:47:15 2009  reduce to 1894462 relation sets and 1860144 unique ideals
Mon Dec 28 18:47:15 2009  ignored 672 oversize relation sets
Mon Dec 28 18:47:15 2009  commencing full merge
Mon Dec 28 18:49:13 2009  memory use: 59.5 MB
Mon Dec 28 18:49:13 2009  found 45455 cycles, need 389341
Mon Dec 28 18:49:13 2009  too few cycles, matrix probably cannot build

[Dec 28 2009, 18:49:14] c123: running 0 auto-increasing ecm curves...

<and after MUCH ECM during christmas holidays, I found this (GMP-ECM started by aliqueit.exe):>

[Mon, Jan. 11, 2010] Run 5415 out of 0:
Using B1=34208163, B2=34208163-144293429296, polynomial Dickson(12), sigma=4281422573
Step 1 took 443469ms
Step 2 took 92906ms
Run 5416 out of 0:
Using B1=34217470, B2=34217470-144293429296, polynomial Dickson(12), sigma=1054942624
^C

So please make msieve give the same "return code" (which is then caught by factMsieve.pl) for both, "filtering wants 1000000 more relations" and the above "error". This would prevent aliqueit.exe from get stuck in the "running ECM forever"-condition (especially useful if aliqueit is run unattended).

[Batalov]

This has been fixed way back in November

[Andi47]

Originally Posted by Batalov

This has been fixed way back in November

In the SVN (Which I didn't know until now, too), but not in the official version. (1.43, which is the latest official version, crashes in this way.)