Chrissie,
patch looks generally good, but do you think that it would be extremely
hard to split it into multiple smaller (compilable) commits (like
changes in coroparse, add cfg_reload_config handler, ...)?
Next important thing is (example):
corosync.conf:
...
crypto_hash: none
...
- exec corosync
- change (in corosync.conf) crypto_hash: none to md5
- reload
- corosync-cmapctl shows crypto_hash: md5, BUT corosync is still using none
This is example of situation, where RO variable is overwritten. There
are multiple examples of these (I've sent you list of them in email "Re:
Corosync.conf reload, RFC" 08/20/2013 02:06 PM). This behavior seems to
be very confusing.
- Now smaller things to fix (please try to send future patches via
git-send-email and not as attachment), format is line number - comment:
80 - Should be 3 in comment
91 - There are spaces instead of tab
115 - Add comment about dependency on sorted ordering. Code STRONGLY
depends on it (what is of course ok) but people forget. So one comment
there will make us not to forget.
204 - Overwriting RO values.
1094 - Please use CMAP_ constants instead of QB_ (yes they are same, but
this may not apply forever)
1192 - Even MESSAGE_REQ_CFG_CRYPTO_SET is unused, we shouldn't renumber
constants
And thats it.
Regards,
Honza
Christine Caulfield napsal(a):
>
>
> _______________________________________________
> discuss mailing list
> discuss@xxxxxxxxxxxx
> http://lists.corosync.org/mailman/listinfo/discuss
>
_______________________________________________
discuss mailing list
discuss@xxxxxxxxxxxx
http://lists.corosync.org/mailman/listinfo/discuss
