[rt-devel] Issue with new users in RT CLI

Iain Price iain.price at post.serco.com
Mon Oct 13 08:56:15 EDT 2003


Iain Price wrote:

> Hi,
>
> For once i'm (hopefully) posting something useful and not just moans 
> about slow databases ;)
>
> I've started implementing the RT CLI tool in my Java API where i was 
> writing direct to the DB before, it seems to work oK although the 
> examples are wrong (there was a patch on a previous mail to fix them), 
> but I had issues with creating new users.  The code seems to create a 
> user called "New user" then modify it to fit ur parameters, the issue 
> (as the code points out) is if this user already exists.  I suspect 
> the comment might refer to the case that the user creates a 'New user' 
> user rather than the code breaking but anyway... if you botch the RT 
> CLI parameters in some way, or something breaks it leaves New user 
> around, and then cant create a new one any more.


Always a bad sign when you reply to yourself :P

Even if this patch isn't used, the default code will proceed to apply ur 
modifications to 'id=NULL' (according to mysql query trace file) and the 
RT cli returns with 'user creation successful', even though it got 
botched rather badly - rather than take my approach assuming New user is 
safe, you could code a error to return here instead (as i initially did 
to debug) - anything would be better than pretending it all went fine :) 
perhaps deeper in the API something should be modified to check if we're 
trying to update ID=NULL and fail, as such a statement surely indicates 
a bug higher up in the code.

Iain





More information about the Rt-devel mailing list