[mapserver-dev] FW: MapServer enhancements/refactoring project

Wed Apr 1 16:20:17 EDT 2009

Tom,

I thought we could have come up with a clear plan before going public 
with this info, and would have thought that a XML and Web Services fan 
like you would have been in support of a move to a better XML and 
service-oriented architecture. Java is robust enough now that you should 
not let the technical details bog you down... think about the future.

This is an opportunity for us to take the MapServer name to the next 
level, based on a more modern architecture and toolset. This will bring 
you the WFS-T and all the OGC services that you've been hoping for. 
Nothing goes away, the C MapServer should continue to be supported by a 
few devs for at least little while, it's just that the next key features 
will be based on the new Java codebase and will be added at a faster 
pace thanks to a reliable source of funding.

Then the MapServer Enterprise gets its real meaning!

Hoping that you'll see the light soon.

Daniel

Kralidis,Tom [Ontario] wrote:
> -dev: see thread below.  I have no choice but to bring in -dev here.
> 
> Daniel: you have got to be kidding me.
> 
> Coming from a non-Java environment, this would represent a major
> showstopper and end of use (and commitment) to MapServer from myself and
> over 100 install bases.  A change as drastic as this will no doubt crush
> confidence in foss4g to those I have been working with so hard to build
> MapServer and foss4g.
> 
> For the last 10 years.
> 
> Thanks a lot!
> 
> Regardless, I think this thread is best discussed with the community who
> should be involved in this discussion.
> 
> ..Tom
> 
> 
> -----Original Message-----
> From: Frank Warmerdam [mailto:warmerdam at pobox.com] 
> Sent: Wednesday, 01 April 2009 15:37
> To: Daniel Morissette
> Cc: Paul Ramsey; thomas bonfort; jmckenna at gatewaygeomatics.com;
> yassefa at dmsolutions.ca; Kralidis,Tom [Ontario]; hobu.inc at gmail.com;
> pnaciona at gmail.com; szekerest at gmail.com; umberto.nicoletti at gmail.com;
> hobu at hobu.net; woodbri at swoodbridge.com
> Subject: Re: MapServer enhancements/refactoring project
> 
> Daniel,
> 
> It's funny that you mention this.  I recently had an expression of
> interest from a client to rewrite GDAL (but not OGR yet) completely in
> Java.  My first reaction was to take a pass but such an effort would be
> a good match with a MapServer rewrite in Java.  If my client will pay
> for the GDAL rewrite into Java, would you be able to set aside a
> man-year or so of funding for me to also rewrite OGR in Java?
> 
> Luckily we can use the GeoTools coordinate system implementation in
> place of PROJ.4, so we shouldn't need to rewrite all of that.
> 
> This level of funded work should take me comfortably through the
> economic meltdown!
> 
> PS. Paul, tell us what you really think!
> 
> Paul Ramsey wrote:
>> I've head enough great potential to last a lifetime. I'll stick with 
>> things that actually work. Re-writing Mapserver in Java is like 
>> tearing down Notre Dame and re-building it in Jello. You and the rest 
>> of the Jello-heads stay away from my church.
>>
>> P.
>>
>> On Apr 1, 2009, at 12:10 PM, Daniel Morissette wrote:
>>
>>> Paul Ramsey wrote:
>>>> Daniel,
>>>> That's quite a big change you're proposing. Not that I have anything
> 
>>>> against Java per se or XML (well, maybe I have something against 
>>>> XML) but I don't want to see either taking over the Mapserver 
>>>> project, money or no money. If it's not written in high-performance,
> 
>>>> highly-painful C I don't want any part of it! Perhaps you ate some 
>>>> bad  > poutine, because the whole idea is off the rocker.
>>> C'mon Paul... of all people I would have thought that you would have 
>>> been the first person to see the great potential in this move!
>>>
>>> As I wrote before, you pick your camp, but others have made the right
> 
>>> decision already.
>>>
>>> Daniel
>>>
>>>
>>> Daniel Morissette wrote:
>>>> The baby has grown, it's an adult now, and getting quite mature, so 
>>>> I wouldn't call that selling the baby but actually letting it 
>>>> explore new grounds.
>>>> Think of it this way: the power and simplicity of MapServer's core 
>>>> algorithms combined with the modern architecture of the Java 
>>>> language and development tools.
>>>> Isn't it about time that we move away from the command-line 
>>>> debuggers and malloc's? It's the 21st century guys... let the IDE 
>>>> and the garbage collector do it all for you. It's been proven 
>>>> several times that the performance is no longer an issue with Java 
>>>> now so why stick with an old language from the 70's when we can do 
>>>> 10 times as much with a modern toolkit?
>>>> We'll probably go ahead whether you're on board or not (I got one 
>>>> private reply from a MapServer dev who says he's on board already), 
>>>> so you choose your camp... continue to pull your hair with the old 
>>>> stuff or have fun taking the MapServer name to the next level with a
> 
>>>> modern architecture.
>>>> Daniel
>>>> thomas bonfort wrote:
>>>>> Daniel,
>>>>>
>>>>> Seeing you are the "mother of MapServer" I would have hoped that 
>>>>> you were not so keen in selling your baby to the first 
>>>>> deep-pocketed java crackhead, even if I now that times are rough 
>>>>> right now and that it's probably more difficult to turn down an 
>>>>> offer that will keep you fed for a confortable amount of time.
>>>>>
>>>>> I would also have preferred that my first vote as a PSC member was 
>>>>> not a veto, as I cannot imagine myself going the java way with
> mapserver.
>>>>> regards,
>>>>>
>>>>> thomas
>>>>>
>>>>> On Wed, Apr 1, 2009 at 20:14, Daniel Morissette 
>>>>> <dmorissette at mapgears.com> wrote:
>>>>>> Guys,
>>>>>>
>>>>>> I've been contacted by a potential client who would be willing to 
>>>>>> fund a complete refactoring of MapServer. It seems that they got 
>>>>>> wind of our discussions at the TO Code Sprint and would like to 
>>>>>> help with some of that, more specifically:
>>>>>> - XML mapfiles
>>>>>> - performance work driven in part by GeoServer which is getting 
>>>>>> very close to MapServer in both performance and map rendering 
>>>>>> quality
>>>>>> - rendering plugin architecture that Thomas is working on
>>>>>> - WFS-T support
>>>>>>
>>>>>> They've got enough budget to get about 20 devs working on this for
> 
>>>>>> a year, but there are two big issues in my mind: they have a tight
> 
>>>>>> deadline, they need this done within a year, and they are 
>>>>>> currently a Java shop so they require that we deliver a Java 
>>>>>> solution (I know, I know). They considered GeoServer but they got 
>>>>>> turned down by the guys at OpenGeo who would not let them get the 
>>>>>> architectural changes they wanted for any amount of money, so they
> 
>>>>>> figured they might have more luck with us. Oh, there is a 3rd
>>>>>> thing:
>>>>>> they would require that the code be closed for at least a year 
>>>>>> after the final deliverable before they let us release anything 
>>>>>> publicly...
>>>>>> something
>>>>>> related to their own internal policies and business plan... not 
>>>>>> ideal but at least in the end the code should be opened up.
>>>>>>
>>>>>> I'm not too keen on a switch to Java, but it seems that the 
>>>>>> performance is no longer an issue as the GeoServer guys have 
>>>>>> demonstrated with their benchmarks, and this would solve many of 
>>>>>> our issues dealing with XML and Web Services (isn't XML the way of
> 
>>>>>> the future anyway?). If the money is there it's hard to turn it 
>>>>>> down, and it could be fun to take over GeoServer on its own 
>>>>>> ground... Java!
>>>>>>
>>>>>> So far I have 4 Java dev's on board. I am writing this to you to 
>>>>>> seek an expression of interest since I'd need at least another 4-5
> 
>>>>>> experienced MapServer devs to undertake this project.
>>>>>>
>>>>>> Who's interested?
>>>>>>
>>>>>> Daniel
>>>>>>
>>
>> --
>> Paul Ramsey
>> pramsey at cleverelephant.ca
>> +1 250 885 0632
>>
>>
> 
> 

-- 
Daniel Morissette
http://www.mapgears.com/