Re: pathological LDM regular-expressions

• To: Tom McDermott <tmcderm@xxxxxxxxxxxxxxxxx>
• Subject: Re: pathological LDM regular-expressions
• From: Steve Emmerson <steve@xxxxxxxxxxxxxxxx>
• Date: Fri, 14 Feb 2003 11:16:01 -0700

Tom,

I'm including the ldm-users mailing-list in my reply because your
question is of general interest.  I hope you don't mind.

>Date: Fri, 14 Feb 2003 13:07:28 -0500 (EST)
>From: Tom McDermott <tmcderm@xxxxxxxxxxxxxxxxx>
>To: Steve Emmerson <steve@xxxxxxxxxxxxxxxx>
>Subject: Re: pathological LDM regular-expressions

The above message contained the following:

> I don't have any REQUEST statements like that in my ldmd.conf, but I do
> have many ALLOW statements with this form:
> 
>       allow ANY               .*.nrcc.cornell.edu
> 
> to allow any host from a particular upstream or downstream institution to
> connect.  Or does this problem only affect REQUEST statements?

This problem is only a major issue with REQUEST entries, so there's no
need to modify the above entry.

> If not,
> what about regular expressions of this form in the pqact.conf file?

It wouldn't hurt to modify them (it might actually speed-up your local
processing). It's just not as important as the REQUEST entries in the
LDM configuration-file.

> Sounds like a bug in the Solaris RE library.

Henry Spencer's regexp(3) library (which is the basis for RE-handling on
several OS-s) also executes the pathological RE-s more slowly (although
not as slowly as Solaris's). It executes the non-pathological RE-s more
slowly than Solaris does, however, so it's a bit of a wash.

Regards,
Steve Emmerson   <http://www.unidata.ucar.edu>