bkbits.net web UI oddities after last crash

bkbits.net web UI oddities after last crash

[Andre Tomt]

Hi

I noticed some rather strange behavior from
http://linux.bkbits.net:8080/linux-2.4

Lets say, I go to
http://linux.bkbits.net:8080/linux-2.4/ChangeSet@-7d?nav=index.html

So far, so good. No oddities here. Then I find "1.1002.3.2" in the list.
"[IPV4]: Be more verbose about invalid ICMPs sent to broadcast."

I click it. Okay, page comes up nicely, the header seems a little
strange, though. But that might just be a feature change (I'm not a
BK-user, yet). It claims I'm looking at ChangeSet
3ef76015xckdLSammmd1kjiJ7CJY4Q. Isn't ChaneSet's named by a number? Not
minding that too much, I click on the "all diffs" link. Returns nothing.
Hmm.. ok, lets click on "diffs" for the single affected file instead,
and behold, the right diff pops up. I try the same on some other changes
in the list, the same problems appear.

Whats happening?

-- 
Cheers,
André Tomt

Re: bkbits.net web UI oddities after last crash

[Larry McVoy]

On Tue, Jun 24, 2003 at 07:27:31PM +0200, Andre Tomt wrote:
> Hi
> 
> I noticed some rather strange behavior from
> http://linux.bkbits.net:8080/linux-2.4

I put up a new release that used MD5 based names for revision numbers rather
than the revision numbers (sort of like naming a file by inode number) 
because the revision numbers shift around on you.  It has bugs.  Once we
work out the bugs the plus is that you'll be able to post a URL and it 
will always get people to the same data which is not true today.
-- 
---
Larry McVoy              lm at bitmover.com          http://www.bitmover.com/lm

Re: bkbits.net web UI oddities after last crash

[Sam Ravnborg]

On Wed, Jun 25, 2003 at 08:46:02AM -0700, Larry McVoy wrote:
> On Tue, Jun 24, 2003 at 07:27:31PM +0200, Andre Tomt wrote:
> > Hi
> > 
> > I noticed some rather strange behavior from
> > http://linux.bkbits.net:8080/linux-2.4
> 
> I put up a new release that used MD5 based names for revision numbers rather
> than the revision numbers (sort of like naming a file by inode number) 
> because the revision numbers shift around on you.

Using cset numbers in the interface would make it a bit easier to
understand.

> It has bugs.  Once we
> work out the bugs the plus is that you'll be able to post a URL and it 
> will always get people to the same data which is not true today.

Good to hear - drop a mail when you think the bugs are fixed so
we know when to give feedback. (Eventually at bitkeeper-users ml)

	Sam