I don't see a motivation to force replacement of tailq, and I don't see a
strong motivation to convert everything to use it (but I don't care).
Matt
On Thu, Mar 28, 2024 at 12:28 PM Frank Filz <ffilzlnx(a)mindspring.com> wrote:
Hello,
Historically, Ganesha has used glist for double linked list. The ntirpc
sub-module uses TAILQ.
Sometime in the past, a TAILQ or two made their way into Ganesha.
Now we have some new contributions that would like to introduce a number
of TAILQ.
From my perspective, having two different list implementations is not
ideal for code maintainability as folks have to remember the subtle
differences between the two kinds of lists.
One question is do we allow the crack of one or two TAILQ to open
floodgates, or do we hold the line and ask new contributions use glist.
The second question, if we do allow more TAILQ is do we then convert all
the glist to TAILQ. Considering we have some other big code hits coming
(binary logging and code format changes) that will make backports a pain,
it would be ideal to convert glist to TAILQ the same release rather that
possibly having the next release also have a significant change that makes
backports a pain. Of course, if we do this conversion, who does it?
If ntirpc wasn’t already using TAILQ, my inclination would be to hold the
line and stick with glist (and maybe go change the one or two TAILQ to
glist).
Thanks for your thoughts,
Frank Filz
_______________________________________________
Devel mailing list -- devel(a)lists.nfs-ganesha.org
To unsubscribe send an email to devel-leave(a)lists.nfs-ganesha.org
--
Matt Benjamin
Red Hat, Inc.
315 West Huron Street, Suite 140A
Ann Arbor, Michigan 48103
http://www.redhat.com/en/technologies/storage
tel. 734-821-5101
fax. 734-769-8938
cel. 734-216-5309