wide-finder Google Group

Using box for concurrency testing in JRuby

2008-11-25T21:15:24Z

If nobody objects I'm going to be doing some concurrency testing in
JRuby for the next couple hours.

- Charlie

Re: Please clean up home dirs on wide-finder

2008-11-12T09:09:21Z

gone

Caoyuan a écrit :

Re: Please clean up home dirs on wide-finder

2008-11-07T11:40:10Z

removed.
On Fri, Nov 7, 2008 at 7:32 PM, Alastair Rankine

Re: Please clean up home dirs on wide-finder

2008-11-07T11:32:58Z

Whoops, sorry. Gone.

Re: Please clean up home dirs on wide-finder

2008-11-07T10:34:14Z

I went ahead and did remove unneeded stuff (a couple build dirs), leaving some
152 MB in my dir, when an idea came to me...
du -sh reports /export/home is taking 23G. Now, there are over 16GB in core
dumps, so simply deleting the following will suffice to recover quite some
space:
11G ./alastair/wf2/core

Please clean up home dirs on wide-finder

2008-11-06T20:49:55Z

/export/home is full and this is causing problems. There's no need to
delete anything, just make a directory on /wf1/data and move your
stuff there, its got tons of space. -Tim

Wednesday, November 05, 2008 8:16 - 9:16 PM

2008-11-05T20:16:25Z

The machine seems free, I'll book the above time.

Thanks.

Re: I'm going to write up Wide Finder, need code

2008-10-20T00:59:17Z

/export/home/erik/WideFinder2. zip
It contains a NetBeans project, so if you use NetBeans and have the Scala
plugin installed, it should be fairly easy to make work. Otherwise it
should be buildable with ant and Scala.

Re: I'm going to write up Wide Finder, need code

2008-10-17T11:29:50Z

These are the versions I made, in chronological order (please find them
attached), along with some comments:
* wf2_simple.ml: naïve port of the reference implementation,
single-core and with line-oriented input. It is an order of magnitude faster
than the original Ruby version.
* wf2_multicore.ml: uses process-based parallelism. This was the first program

Re: I'm going to write up Wide Finder, need code

2008-10-16T13:12:47Z

Tim,
I'll get my code posted this weekend.
Incidentally, this thread got marked as spam by GMail. Wonder why...
-Erik

Re: I'm going to write up Wide Finder, need code

2008-10-16T08:56:47Z

I managed to corrupt my pointers, I meant to reference the message
I got about awka there:

[link]

Re: I'm going to write up Wide Finder, need code

2008-10-16T08:49:52Z

git clone [link]
To get to any arbitrary revision (especially those I mentioned in posts here)
in my repo:
git reset --hard <sha1>
I think my favorite[1] was with:
e80624c7513c15a4d6154fd4c0c189 27cd5b1b5c
In case you're still having trouble, I've also packaged a tarball here:

Re: I'm going to write up Wide Finder, need code

2008-10-15T23:31:34Z

This is certainly what happened to me - for some reason Alastair's got
through however, which prompted me to go looking...
There isn't much I'd highlight, except that I got stuck with Python on
a combination of incomplete support for OS-level things like mmap()
and the complexity of identifying where my actual bottlenecks were (as

Re: I'm going to write up Wide Finder, need code

2008-10-15T23:02:26Z

Tim,
To answer your question, my code is in /export/home/alastair/wf2/wf2. cpp
I too was planning to give a talk on my experience with WF2 but canned
it, mainly due to positive result bias on my part.
In summary though, my experience with WF2 was one of mostly incorrectly
guessed bottlenecks.
My initial implementations of WF2 were developed using the smaller 1m

Re: I'm going to write up Wide Finder, need code

2008-10-15T20:15:55Z

BTW, I found your email in my gmail spam bin. I think the recent series of
spam on the widefinder mailing list might be causing similar problems for
other folks. Maybe that explains the lack of response so far...
-Ray