Dave Cecil wrote:
>
> Daniel,
>
> >Here's the proposed mount-time tailmerging algorithm which was originally
> >posted as an attachment summarizing a series of emails between me and Stephen
> >Tweedie on this subject:
> >-------------------------------
>
> [snip]
>
> first of all I admit to only just joining this list and consequently, I
> don't know the history of this thread, so I'd ask that you forgive me if I
> now ask a stupid question ;-) Just tell me to pull my head in.
>
> Would this algorithm be clearer to all interested parties if you had a
> couple of diagrams? I guess you only want to do this if you want some peer
> review by those who are "visual" people. If you're just politely
> communicating what you and Steven have already decided, then fair
> enough. It's just that I struggle to see what's happening without a
> diagram or two. Maybe something on a web page? Even some scanned sketches...
It's a throwaway algorithm, something like a scaffold for working on a
building. Its purpose is to get something up and running, then it will
be replaced by an as-yet-undesigned incremental algorithm. If it lives
longer than that for some reason, it would be worth making a diagram.
On the other hand, a diagram of the way tail blocks are shared would be
well worth making - why not read the thread(s):
http://web.gnu.walfield.org/mail-archive/linux-fsdevel/
and email your diagram to:
[EMAIL PROTECTED]
--
Daniel
