[Mongrel] Invalid HTTP format, parsing fails

Ross Singer ross.singer at library.gatech.edu
Thu Aug 24 18:27:55 EDT 2006


Hi,

I am having a bit of a problem with my app that is using mongrel.

For a particular vendor we use, they are generating links that look like:
http://umlaut.library.gatech.edu/resolve?genre=article&issn=00224898&title=Journal+of+Terramechanics&volume=43&issue=4&date=20061001&atitle=Requirements+and+system+design+for+a+robot+performing+selective+cleaning+in+young+forest+stands.&spage=505&sid=EBSCO:aph&pid=Vestlund%2c+Karin%3bHellstr%c3%b6m%2c+Thomas
>>2183092720061001aph

(pardon the awfulness of OpenURL query strings)

Which for IE gives the user:
Thu Aug 24 18:12:48 EDT 2006: BAD CLIENT (127.0.0.1): Invalid HTTP format,
parsing fails.

I am not entirely sure, but I think too many of these eventually bring
mongrel down completely (mongrel keeps stopping, it may or may not be
related).

Firefox and its ilk properly escape these query strings, so those users
aren't affected by it (except, of course, when mongrel goes down
completely).

Removing the ">>" eliminates the problem in IE.

I've placed a call to the vendor in question to please stop doing this, but
I don't know if I'll get any response, and, besides, it's quite possible
I'll see more invalid characters from other sources in the future.

Webrick also fails in exactly the same way on the same queries.  FastCGI
doesn't (as far as I can tell), but I also haven't had any luck getting
FastCGI working with this app.

I'm using mongrel 0.3.13.3 in production and 0.3.13.4 in development (the
link above).

Thanks for any help you can offer on this,
-Ross.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://rubyforge.org/pipermail/mongrel-users/attachments/20060824/6d316ecb/attachment.html 


More information about the Mongrel-users mailing list