[mapserver-dev] 6.0 release plan

[Tamas Szekeres]

2010/11/8 Lime, Steve D (DNR) <Steve.Lime at state.mn.us>

>
> RFC-52: The changes I’d like to see are relatively minor (from a coding
> perspective) and are a return to a single getShape() method in MapScript. It
> will break 5.6 scripts but is simpler in the long run. I don’t know that an
> RFC is necessary but I will start a ticket and we can go from there.
>

Steve,

Upon thinking about the one-pass query implementation for the mssql2008
driver I would also prefer to establish an additional parameter in
msLayerWhichShapes which would denote the purpose of the query. Actually the
driver may require to use a different fetching method if only a single-row
forward only cursor is sufficient (which may be faster) or we would require
to have subsequent random access to the result set. The latter would require
to open the the query with the block cursor option at the client side which
may be less performant for larger rowset sizes.



>
>
> RFC-64: Needs feedback. This may push us back a week or two depending on
> feedback. I think 6.0 is the right time to implement though given the types
> of proposed changes. They are not candidates for minor releases.
>
>
>

While I'm not sure about the recent implementation, I would highly support
the efforts in RFC-64 in the way as Frank suggested earlier. In this case
instead of using globals to represent the state, we could probably use the
parse-param and lex-param directives to have a 'context' data structure
something like in
http://trac.osgeo.org/gdal/browser/sandbox/warmerdam/gdal-rfc28/gdal/ogr/swq_parser.y
used for GDAL. This would make the parser thread safe without requiring a
global lock around the operations and I would also expect a significant
increment in the overall performance when rendering maps in multithreaded
environments.


Best regards,

Tamas
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.osgeo.org/pipermail/mapserver-dev/attachments/20101108/aebbb50e/attachment.html