agg changes

[Steve Lime]

>>> On 8/28/2007 at 3:50 PM, in message
<d2b988930708281350s5644a102yd18348e52c549bfe at mail.gmail.com>, thomas bonfort
<thomas.bonfort at GMAIL.COM> wrote:
> On 8/28/07, Steve Lime <Steve.Lime at dnr.state.mn.us> wrote:
>>
>> I think there will be a number of AGG specific changes we might want to
>> look at. Arrowheads are one
>> other example. I can extend styles for 5.2 if we desire to do that.
> 
> 
> as well as a few 'native' symbols maybe, without the need to specify them
> explicitely in the mapfile / symbols.sym?

There actually a but out there about defining named (i.e. native) symbols that would serve
this purpose. We'd just pre-populate the symbol set with and access them with special names
like _triangle or something like that. The defs would be complete symbols so GD could work with
them but the AGG code could ignore the symbol and just use the name to trigger use of the 
built-in markers it supports.

> Cap/join are only used for cartolines (that I could see) so if that option
>> were to go away they could too.
>> With AGG there's really no reason to use 'em.
> 
> 
> they can be used in agg too... miter and bevel joins could be usefull, as
> well as butt and square caps... maybe not a used/usefull enough feature to
> support though, I don't know about that

I mean't to use cartolines. AGG gives you everything we were trying to provide with cartolines
but only better. So why keep them. Joins/caps should continue to exist but in the style obj.

> Since the AGG support is so strong I would like to consider removing
>> non-paletted support for GD
>> altogether, that is no RGB or RGBA GD drivers. That would really simplify
>> life and provide the most
>> consistent output in each case. We'd be left with:
>>
>>   - RGB/RGBA coming from AGG
>>   - 256 color output from GD or AGG with a palette or quantization enabled
>>
>>
>> Each library would be used for what it does best. Perhaps too radical?
> 
> perhaps too radical I think... I'd prefer effort going into having renderers
> presenting a common interface to mapserver, but that's  maybe further off...

With such an interface I would only implement PC256 GD support going forward... 

> Steve
>>
>> >>> On 8/28/2007 at 3:58 AM, in message
>> < d2b988930708280158g4e195a24g37918523ad242687 at mail.gmail.com>, thomas
>> bonfort
>> <thomas.bonfort at GMAIL.COM> wrote:
>> > On 8/27/07, Steve Lime < Steve.Lime at dnr.state.mn.us> wrote:
>> >>
>> >> I would agree with Zak, round for both (mimic the circular brush in
>> GD).
>> >> We'll have to move those
>> >> keywords out of symbolObj's and into styleObj's.
>> >
>> >
>> > just a few remarks here...
>> > removing cartolines is fine when using AGG, but gd still uses them if
>> not
>> >
>> > default caps are now round again.
>> >
>> > I *don't* think we should remove the caps/join keywords from the
>> symbolobj,
>> > as they are used when drawing vector symbols (for fills, markers, and
>> > polyline markers from the top of my head).
>> > I *do* think having the same keywords in the stylobj is a good idea. I
>> don't
>> > think I want to go diving in the lexer code etc.. to add this myself,
>> but I
>> > can quickly add this to the line rendering code once this is settled out
>> >
>> > thomas
>>
>>