Custom Query (1630 matches)

This should be forgotten after the commit has completed. Look into the docs why it's remembering this.

There seem to be scheduler problems with Spiffy on call-cc.org, as it has hung up several times, now.

Suggested plan:

• try to reproduce it by hammering the server locally with ab or Siege.

• checkout the scheduler patches submitted recently by Jörg Wittenberger (spiffy on call-cc.org runs on 4.5.0, the diff is attached).

• possibly try Joerg's "high-load-scheduler" (needs to be updated to the newest version, first, since JW sent me some code recently that introduced some changes that haven't been integrated).

• Stay calm.

Hello esteemed team leader.

I am having a reproducible error pulling down the sql-de-lite egg from either mirror, which I have tracked down to commit 7bd92ff78681ddee949441114bcc8bd6bd6a4284. The symptom is below:

$ chicken-install -n -l http://code.call-cc.org/cgi-bin/henrietta.cgi sql-de-lite
retrieving ...
connecting to host "code.call-cc.org", port 80 ...
requesting "/cgi-bin/henrietta.cgi?name=sql-de-lite&mode=default" ...
reading response ...
HTTP/1.1 200 OK
Date: Thu, 12 Aug 2010 04:04:47 GMT
Server: Apache/2.2.9 (Debian) DAV/2 SVN/1.5.1 mod_ssl/2.2.9 OpenSSL/0.9.8g
Connection: close
Vary: Accept-Encoding
Transfer-Encoding: chunked
Content-Type: text/plain
reading chunks ...

[pause of about 6 seconds]

Error: (=) bad argument type - not a number: #f

Something in that commit 7bd92ff78 is causing the error; the parent commit works. I have to assume it is an interaction between TCP and the new flonum milliseconds stuff. I am using experimental HEAD (c6d6f125c1eed52dd9e8540fc4454eea95472b09) which still has this issue. Some other eggs I've tried seem to work, so perhaps this is because sql-de-lite is transferring 4MB.

Unfortunately I haven't tracked down the exact cause yet, but hopefully this will help.