Header parsing in 2.0a15,a18?

I've got some, uh, rather large email folders that I've been trying to convert to html with hypermail. I've been tyring the newer alpha releases because of the MIME support, but see lots of problems with header parsing. The results vary with the type/size of folder, but I see:

• segmentation faults part-way through the mail (one case is on email message 340, another folder has if on message 94)
• quietly dropped email - one 2.6MB folder with 700+ messages results (quite nicely, however, no seg fault) in 3 email messages being extracted and converted - these are not the first three, or last three, but three messages from the middle of the file - they are not consecutive messages either. Another (much smaller, with only 9 email messages) file calmly produces no email extracted.

I've applied the patch to 2.0a18 to remove the spurious "level" argument in the threadprint.c file, incidently. 2.0a15 shows similar (identical, in fact) problems.

Now, just to confuse things, 2.0a3 has no problem extracting the messages correctly.

What these folders all have in common is that they are folders holding copies of *outgoing* mail. I don't think I've seen the problem on folders on incoming mail (though none of those folders are as big on my system).

Is anyone else seeing this?

--
Steve Wampler-  SOLIS Project, National Solar Observatory
swampler_at_noao.edu