ReadSeq large sequence bug

[Francis Ouellette]

gilbertd at sunflower.bio.indiana.edu (Don Gilbert) writes:

>There is a bug in the current release of readseq (dated 1Feb93)
>which causes it to abort when fed sequences larger than a certain
>size, at least when compiled with certain compilers.  If you
>feed a sequence of around 5000 base pairs or larger to readseq,
>and it aborts (maybe with the message "Segmentation fault") then
>you have found the bug.

>Here is a temporary correction for that bug.
>In the file ureadseq.c, change near top:

>#define kStartLength  500
>	to:

>#define kStartLength  500000   /* 20Apr93 temp. bug fix */
>	/* set this value higher than max seq size expected */

>then recompile it, and readseq will work on large sequences. 
>(tobacco chlorop. gene  acc#Z00044 with  155844 bp for example)

>But this isn't a definative bug fix.  It looks like the
>memory management code for calloc/malloc/realloc is failing
>where it shouldn't.   I will have to look at it more.


well it works fine on 50-60 k files, but it did core dump with 
the yeast chromosome III file (315k)   :-)

(on a sparc2, SunOS 4.1.2, compilled with gcc (I forget the version #)

but this is fine (for now anyways!)

thank you very much Don,

francis
--
| B.F. Francis Ouellette     * francis at monod.biol.mcgill.ca *
|
|      "Je cherche a` comprendre"  Jacques Monod