On 2006-08-11T16:02:09, David Lee <[EMAIL PROTECTED]> wrote:
> +1 for the rip-out. CVS (and its successors) should be robust enough for
> any occasional necessary "oops! un-rip" operations.
>
> Perhaps you should begin this almost as soon as 2.0.7 ships. Then folk
> here on this list will have plenty of time to stumble across any of those
> occasional "un-rip" requirements.
I'm delaying major mongery right now because Andrew is >< close with the
mercurial conversion...
> I suspect that when the first, rough draft of a new, experimental module
> goes in, it doesn't get seriously peer-reviewed. That's understandable...
Ah, no, not really. The design needs to be reviewed at exactly this
stage. Implementation details can always be fixed and are more easy to
refactor.
(Design issues can be fixed too of course, but it's way more work - ask
Andrew about the estimated effort for the PE rewrite ;-)
> it is a first, rough draft. Any peer review tends to kick in only later,
> by which time any overall poor design (include opaque structure and
> spaghetti code) is too deeply embedded, and that peer-review is only
> focussed on the particular detail, not the "big picture".
No, quite honestly, everyone was too busy one way or the other to review
code going in early, I think. We simply need to get better at this. We
need some Christopher Hellwigs on our project ;-)
> OK. I had recently been attempting to look into "lrmd" to understand a
> problem and finding it impossible to get to grips with. So, from a purely
> programming-style view point, I had suggested this as a candidate for
> possible re-write into a completely different language, as it just might
> possibly result in a better product. But the additional considerations
> you mention almost certainly block that idea.
Well, I feel like that with the PE sometimes too ;-) Some time to get
started on something is certainly sound. But, the PE is vastly more
complex, and despite the typos, Andrew's comments don't mislead me as
often.
And, with sunjd now having moved on to new challenges, we need someone
to look after the lrmd & stonithd. I guess they are as good a place as
any for me to make some time for actual coding again. Probably after my
vacation next week, unless I get too bored.
(Also, Andrew can then complain about lrmd/stonithd to his boss, that
will improve the style & mood in which these comments are made I'm sure
;-)
Sincerely,
Lars
--
High Availability & Clustering
SUSE Labs, Research and Development
SUSE LINUX Products GmbH - A Novell Business -- Charles Darwin
"Ignorance more frequently begets confidence than does knowledge"
_______________________________________________________
Linux-HA-Dev: [email protected]
http://lists.linux-ha.org/mailman/listinfo/linux-ha-dev
Home Page: http://linux-ha.org/
