Makehistory, qio issues

Russ Allbery rra at stanford.edu
Thu Jun 10 20:34:49 UTC 2004


Heath Kehoe <hakehoe at avalon.net> writes:

> Hmm... I see a fix for dealing with too-long lines is in CURRENT (qio.c
> 1.11). Has that fix been sufficiently tested to back-port to STABLE?

I think so, yes.  QIO has a pretty comprehensive test suite in CURRENT.
You should be able to backport that fix safely.

(Oh, btw, when committing the same change to multiple branches, could you
just copy/paste the commit message?  Commit messages that refer to other
changes by revision number may end up losing their meaning if we ever
switch to some other revision system.)

> Also, I see that the qio code uses the blocksize of the filesystem as
> its buffer size (unless it's larger than 4x or smaller than 1/2 of
> QIO_BUFFERSIZE) which in my case is 4096.

> I would like to make the following change to qio.c, which makes sure
> that the buffer size will always be at least 8192, but still be a
> multiple of st_blksize.  Any objections?

None here.

-- 
Russ Allbery (rra at stanford.edu)             <http://www.eyrie.org/~eagle/>

    Please send questions to the list rather than mailing me directly.
     <http://www.eyrie.org/~eagle/faqs/questions.html> explains why.


More information about the inn-workers mailing list