Automated tests (was Call for 4.8.3 release...)

Stephen Woodbridge woodbri at SWOODBRIDGE.COM
Mon Mar 27 19:27:50 EST 2006


Steve, et al,

As a user of mapserver, my policy has always been to wait for few dot 
releases before considering a new major release as being stable enough 
for production evaluation. I think that about says most it right there.

I think the work that Sean has done has improved the situation 
significantly. I personally would like to see more testing and I was 
hoping that the unit test efforts would have been more openly accepted 
and supported in some form and that they would move forward. That effort 
has gone quiet, has it died?

It would be good if the project would get behind any developer that is 
will to put forth the effort to expand our testing even to the extent 
that it might require some small changes in the development process. I 
understand that making big development cultural changes around an 
existing project is next to impossible, but more support along the lines 
of build it, we will help and we will use it would be great.

A couple of ways of proceeding would be to look into test coverage 
analysis if anyone has any experience with that. Another would be to 
actively recruit someone to focus on test suite development that could 
work on a rotating basis with existing developers to backfill tests. 
Another would be a policy of adding a section to the RFC that requires 
developer of new features to identify the features and functionality and 
how they plan to extend the automated testing to cover that 
functionality and to state what coverage they do not plan to test.

These are all ideas that are fodder from the mill, so feel free to grind 
them up. I hope some might be helpful.

-Steve W.


Steve Lime wrote:
> This brings us to another point. Sean's tests identified the
> embedding problem but were not run by me (or anyone else as far as I
> know) prior to the 4.8.2 release. In fact that bug probably was there
> at the 4.8.0 release. Others (Sean, Umberto and so on) have pushed
> for more testing and this issue underscores the fact. However, IMHO
> the tests need to be part of the "build" process so a make test
> and/or make pytest could trigger things very easily.
> 
> This is a hot topic on the OSGeo incubator list. How should we
> proceed as a project?
> 
> Steve
> 
> 
>>>> Frank Warmerdam <warmerdam at POBOX.COM> 03/27/06 2:56 PM >>>
> 
> Steve Lime wrote:
> 
>> Hi all: We fixed a somewhat obscure but nasty bug with embeded
>> legends and scalebars over the weekend. Basically you can't use
>> that feature in 4.8 at all. It might make sense to cut a 4.8.3 to
>> mitigate the problem. Looks like there was also a PostGIS patch
>> that wasn't applied to 4.8.2. Any objections?
>> 
>> Steve
> 
> 
> Steve,
> 
> Please!  With all speed!
> 
> Best regards,



More information about the mapserver-dev mailing list