Revised 4.8 release plan

[Sean Gillies]

On Dec 12, 2005, at 4:15 PM, Daniel Morissette wrote:

> With all the chaos of the last few weeks, we are a bit behind relative
> to our original 4.8 release schedule: we should have had our release
> candidate last Friday and our final release this Friday.
>
> We may have to delay the release by a few days, but I would rather not
> delay it later than Dec 23rd because after that we get into the  
> holidays
> and it becomes really hard to get a hold of people. I also know  
> that the
> new website team has been working real hard to try to get things ready
> in time for our initial planned date, so I would not want to ruin  
> their
> efforts.
>
> So I would like to propose the following revised plan:
>
> 4.8.0-rc1 on Thursday Dec 15th
> 4.8.0 final on Thursday Dec 22nd
>
> Are there major outstanding bugs that you were actively working on and
> that would prevent us from proceeding with those dates?
>
> Also, Steve is working on a "fuzzy brush" last minute addition that
> would allow producing antialiased thick lines and he would like to get
> it into this release. I know that this is clearly a new feature and
> should not be allowed after the feature freeze, but the addition is
> relatively straightforward and would constitute a major enhancement  
> for
> this release. Are there any objections to exceptionally breaking the
> rule and allowing this last minute addition knowing that it is a very
> localized change and has very little chance of breaking anything else
> (just a new function to create fuzzy brushes and probably one new
> mapfile parameter)?
>
> Steve: would you be able to get your change in by Wednesday?
>
> Daniel
> -- 
> ------------------------------------------------------------
>   Daniel Morissette               dmorissette at dmsolutions.ca
>   DM Solutions Group              http://www.dmsolutions.ca/
> ------------------------------------------------------------

Daniel,

I'm OK with the revision, but have some reservations about the fuzzy  
brush feature. Adding a new mapfile parameter can be disruptive  
because then you've gotta update the swig docs, update the mapfile  
save/writing code, update msCopyStyle, etc. This weekend I found that  
msCopyStyle currently segfaults in a test that has been passing for  
several releases, probably due to some subtle change in style or  
class initialization. There are also a number of new memory leaks in  
and around classObj.

On the other hand, it does seem like a good feature. I'm for it if  
all the side effects are accounted for.

Sean