On Thu, 17 May 2001, Daniel Phillips wrote:
> Well, if you look how I did the index, it works with blocks and buffers
> while still staying entirely in the page cache. This was Stephen's
> suggestion, and it integrates reliably with Al's page-oriented code.
> So I'm mixing pages and blocks together and it's working pretty well.
... or, in immortal words of Hans, "Yura, run the benchmarks"...
> BTW, the parts of Al's patch that I converted from pages to blocks got
> shorter and easier to read.
<LART>
No offense, but your code is a twisted mess to the degree when
bugs are hard to see just because of obfuscation.
Please, find and describe an obvious bug in ext2_add_entry()
(dx-2.4.4-6.pcache version). After deobfuscation it becomes immediately
visible - the only reason why it doesn't stick out like a sore tumb is
that code is a spaghettish mess.
Generally, when one has to draw a flowchart to figure out what
happens in a function and to find lifetimes of local variables (couple of
dozens of them) it means only one thing: function is _crap_.
And yes, flowchart is what I finally had to resort to. Daniel, I
don't care if you consider writing hairy code as a DSW and frankly, I'm
less than impressed by the S of D being demonstrated in that particular
case.
</LART>
Al
-
To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in
the body of a message to [EMAIL PROTECTED]