Tile caching

[JasonBirch]

Hi...

I was just looking at Zac's recent ticket about adding cache headers to
GETTILEIMAGE requests.

https://trac.osgeo.org/mapguide/ticket/467

I don't think that this is the best way of presenting our tile caches.
Personally, I'd prefer to see a direct HTTP interface to the cache,
where the user requests a file from the webserver as a resource.  That
way all of the HTTP-goodness (cacheing, etc) is built-in.

I don't think that this would be all that hard to do.  There would have
to be a 404 handler that either generates and sends back images that
don't exist or, if it is unable to generate the image send the
appropriate 400- or 500-level codes.  The only problem that I can see
with this is security.  There would have to be some way for either
MapGuide to use OS or LDAP security (and the web server to use the same)
or for the web server to use the MapGuide user store for authentication.

There was some discussion about MapGuide security earlier (at FOSS4G?)
and I certainly wouldn't be sad to see MapGuide's proprietary security
model replaced with something that tied into OS-level security.

Thoughts?

Jason
_______________________________________________
mapguide-internals mailing list
mapguide-internals@...
http://lists.osgeo.org/mailman/listinfo/mapguide-internals

[zspitzer]

On Fri, Feb 29, 2008 at 4:48 AM, Jason Birch <Jason.Birch@...> wrote:

> Hi...
>
>  I was just looking at Zac's recent ticket about adding cache headers to
>  GETTILEIMAGE requests.
>
>  https://trac.osgeo.org/mapguide/ticket/467
>
>  I don't think that this is the best way of presenting our tile caches.
>  Personally, I'd prefer to see a direct HTTP interface to the cache,
>  where the user requests a file from the webserver as a resource.  That
>  way all of the HTTP-goodness (cacheing, etc) is built-in.

Tilecache http://tilecache.org/ :)

Writing a mapguide rendering backend plugin for tilecache would
probably be the easiest effort wise. It
would require adding python into the mix, but it's a code base that is
already written and working
in production.

I think going if any changes in this area should adopt existing
standards or adopt existing patterns
where possible (GETTILEIMAGE and WMS feel a bit like indentical twins
separated at birth).

>
>  I don't think that this would be all that hard to do.  There would have
>  to be a 404 handler that either generates and sends back images that
>  don't exist or, if it is unable to generate the image send the
>  appropriate 400- or 500-level codes.  The only problem that I can see
>  with this is security.  There would have to be some way for either
>  MapGuide to use OS or LDAP security (and the web server to use the same)
>  or for the web server to use the MapGuide user store for authentication.
>
>  There was some discussion about MapGuide security earlier (at FOSS4G?)
>  and I certainly wouldn't be sad to see MapGuide's proprietary security
>  model replaced with something that tied into OS-level security.

>
>  Thoughts?
>
>  Jason
>  _______________________________________________
>  mapguide-internals mailing list
>  mapguide-internals@...
>  http://lists.osgeo.org/mailman/listinfo/mapguide-internals
>



--
Zac Spitzer -
http://zacster.blogspot.com (My Blog)
[phone]
_______________________________________________
mapguide-internals mailing list
mapguide-internals@...
http://lists.osgeo.org/mailman/listinfo/mapguide-internals

[Traian Stanev]

I don't know much about tilecache and WMS-C -- is that the same format as the NASA tiled WMS?

http://onearth.jpl.nasa.gov/tiled.html


Traian

________________________________________
From: mapguide-internals-bounces@... [mapguide-internals-bounces@...] On Behalf Of Zac Spitzer [zac.spitzer@...]
Sent: Thursday, February 28, 2008 7:32 PM
To: MapGuide Internals Mail List
Subject: Re: [mapguide-internals] Tile caching

On Fri, Feb 29, 2008 at 4:48 AM, Jason Birch <Jason.Birch@...> wrote:

> Hi...
>
>  I was just looking at Zac's recent ticket about adding cache headers to
>  GETTILEIMAGE requests.
>
>  https://trac.osgeo.org/mapguide/ticket/467
>
>  I don't think that this is the best way of presenting our tile caches.
>  Personally, I'd prefer to see a direct HTTP interface to the cache,
>  where the user requests a file from the webserver as a resource.  That
>  way all of the HTTP-goodness (cacheing, etc) is built-in.

Tilecache http://tilecache.org/ :)

Writing a mapguide rendering backend plugin for tilecache would
probably be the easiest effort wise. It
would require adding python into the mix, but it's a code base that is
already written and working
in production.

I think going if any changes in this area should adopt existing
standards or adopt existing patterns
where possible (GETTILEIMAGE and WMS feel a bit like indentical twins
separated at birth).

>
>  I don't think that this would be all that hard to do.  There would have
>  to be a 404 handler that either generates and sends back images that
>  don't exist or, if it is unable to generate the image send the
>  appropriate 400- or 500-level codes.  The only problem that I can see
>  with this is security.  There would have to be some way for either
>  MapGuide to use OS or LDAP security (and the web server to use the same)
>  or for the web server to use the MapGuide user store for authentication.
>
>  There was some discussion about MapGuide security earlier (at FOSS4G?)
>  and I certainly wouldn't be sad to see MapGuide's proprietary security
>  model replaced with something that tied into OS-level security.

>
>  Thoughts?
>
>  Jason
>  _______________________________________________
>  mapguide-internals mailing list
>  mapguide-internals@...
>  http://lists.osgeo.org/mailman/listinfo/mapguide-internals
>



--
Zac Spitzer -
http://zacster.blogspot.com (My Blog)
[phone]
_______________________________________________
mapguide-internals mailing list
mapguide-internals@...
http://lists.osgeo.org/mailman/listinfo/mapguide-internals
_______________________________________________
mapguide-internals mailing list
mapguide-internals@...
http://lists.osgeo.org/mailman/listinfo/mapguide-internals

[zspitzer]

On Fri, Feb 29, 2008 at 11:36 AM, Traian Stanev
<traian.stanev@...> wrote:
>
>  I don't know much about tilecache and WMS-C -- is that the same format as the NASA tiled WMS?
>
>  http://onearth.jpl.nasa.gov/tiled.html

Pretty much i think but you can teach tilecache about different backends,

the existing tilecache providers live here
http://svn.tilecache.org/trunk/tilecache/TileCache/Layers/

refactoring Mike Adair's openlayers mapguide support into python and
adding auto configuration
of the tile grid using getresourcecontent would be quite straight forward

>
>
>  Traian
>
>  ________________________________________
>  From: mapguide-internals-bounces@... [mapguide-internals-bounces@...] On Behalf Of Zac Spitzer [zac.spitzer@...]
>  Sent: Thursday, February 28, 2008 7:32 PM
>  To: MapGuide Internals Mail List
>  Subject: Re: [mapguide-internals] Tile caching
>
>
>
>  On Fri, Feb 29, 2008 at 4:48 AM, Jason Birch <Jason.Birch@...> wrote:
>  > Hi...
>  >
>  >  I was just looking at Zac's recent ticket about adding cache headers to
>  >  GETTILEIMAGE requests.
>  >
>  >  https://trac.osgeo.org/mapguide/ticket/467
>  >
>  >  I don't think that this is the best way of presenting our tile caches.
>  >  Personally, I'd prefer to see a direct HTTP interface to the cache,
>  >  where the user requests a file from the webserver as a resource.  That
>  >  way all of the HTTP-goodness (cacheing, etc) is built-in.
>
>  Tilecache http://tilecache.org/ :)
>
>  Writing a mapguide rendering backend plugin for tilecache would
>  probably be the easiest effort wise. It
>  would require adding python into the mix, but it's a code base that is
>  already written and working
>  in production.
>
>  I think going if any changes in this area should adopt existing
>  standards or adopt existing patterns
>  where possible (GETTILEIMAGE and WMS feel a bit like indentical twins
>  separated at birth).
>
>  >
>  >  I don't think that this would be all that hard to do.  There would have
>  >  to be a 404 handler that either generates and sends back images that
>  >  don't exist or, if it is unable to generate the image send the
>  >  appropriate 400- or 500-level codes.  The only problem that I can see
>  >  with this is security.  There would have to be some way for either
>  >  MapGuide to use OS or LDAP security (and the web server to use the same)
>  >  or for the web server to use the MapGuide user store for authentication.
>  >
>  >  There was some discussion about MapGuide security earlier (at FOSS4G?)
>  >  and I certainly wouldn't be sad to see MapGuide's proprietary security
>  >  model replaced with something that tied into OS-level security.
>
>
>  >
>  >  Thoughts?
>  >
>  >  Jason
>  >  _______________________________________________
>  >  mapguide-internals mailing list
>  >  mapguide-internals@...
>  >  http://lists.osgeo.org/mailman/listinfo/mapguide-internals
>  >
>
>
>
>  --
>  Zac Spitzer -
>  http://zacster.blogspot.com (My Blog)
>  [phone]
>  _______________________________________________
>  mapguide-internals mailing list
>  mapguide-internals@...
>  http://lists.osgeo.org/mailman/listinfo/mapguide-internals
>  _______________________________________________
>  mapguide-internals mailing list
>  mapguide-internals@...
>  http://lists.osgeo.org/mailman/listinfo/mapguide-internals
>

--
Zac Spitzer -
http://zacster.blogspot.com (My Blog)
[phone]
_______________________________________________
mapguide-internals mailing list
mapguide-internals@...
http://lists.osgeo.org/mailman/listinfo/mapguide-internals

[zspitzer]

Geowebcache looks like an interesting option in this area

http://google-opensource.blogspot.com/2008/05/accelerate-your-maps-with-geowebcache.html

the actual project website has been down for a while tho

but in order for this to be useful, Mapguide would need to expose
mapDefinitions via WMS, not just layers

z

On Fri, Feb 29, 2008 at 11:08 AM, Zac Spitzer <zac.spitzer@...> wrote:

> On Fri, Feb 29, 2008 at 11:36 AM, Traian Stanev
> <traian.stanev@...> wrote:
>>
>>  I don't know much about tilecache and WMS-C -- is that the same format as the NASA tiled WMS?
>>
>>  http://onearth.jpl.nasa.gov/tiled.html
>
> Pretty much i think but you can teach tilecache about different backends,
>
> the existing tilecache providers live here
> http://svn.tilecache.org/trunk/tilecache/TileCache/Layers/
>
> refactoring Mike Adair's openlayers mapguide support into python and
> adding auto configuration
> of the tile grid using getresourcecontent would be quite straight forward
>
>
>>
>>
>>  Traian
>>
>>  ________________________________________
>>  From: mapguide-internals-bounces@... [mapguide-internals-bounces@...] On Behalf Of Zac Spitzer [zac.spitzer@...]
>>  Sent: Thursday, February 28, 2008 7:32 PM
>>  To: MapGuide Internals Mail List
>>  Subject: Re: [mapguide-internals] Tile caching
>>
>>
>>
>>  On Fri, Feb 29, 2008 at 4:48 AM, Jason Birch <Jason.Birch@...> wrote:
>>  > Hi...
>>  >
>>  >  I was just looking at Zac's recent ticket about adding cache headers to
>>  >  GETTILEIMAGE requests.
>>  >
>>  >  https://trac.osgeo.org/mapguide/ticket/467
>>  >
>>  >  I don't think that this is the best way of presenting our tile caches.
>>  >  Personally, I'd prefer to see a direct HTTP interface to the cache,
>>  >  where the user requests a file from the webserver as a resource.  That
>>  >  way all of the HTTP-goodness (cacheing, etc) is built-in.
>>
>>  Tilecache http://tilecache.org/ :)
>>
>>  Writing a mapguide rendering backend plugin for tilecache would
>>  probably be the easiest effort wise. It
>>  would require adding python into the mix, but it's a code base that is
>>  already written and working
>>  in production.
>>
>>  I think going if any changes in this area should adopt existing
>>  standards or adopt existing patterns
>>  where possible (GETTILEIMAGE and WMS feel a bit like indentical twins
>>  separated at birth).
>>
>>  >
>>  >  I don't think that this would be all that hard to do.  There would have
>>  >  to be a 404 handler that either generates and sends back images that
>>  >  don't exist or, if it is unable to generate the image send the
>>  >  appropriate 400- or 500-level codes.  The only problem that I can see
>>  >  with this is security.  There would have to be some way for either
>>  >  MapGuide to use OS or LDAP security (and the web server to use the same)
>>  >  or for the web server to use the MapGuide user store for authentication.
>>  >
>>  >  There was some discussion about MapGuide security earlier (at FOSS4G?)
>>  >  and I certainly wouldn't be sad to see MapGuide's proprietary security
>>  >  model replaced with something that tied into OS-level security.
>>
>>
>>  >
>>  >  Thoughts?
>>  >
>>  >  Jason
>>  >  _______________________________________________
>>  >  mapguide-internals mailing list
>>  >  mapguide-internals@...
>>  >  http://lists.osgeo.org/mailman/listinfo/mapguide-internals
>>  >
>>
>>
>>
>>  --
>>  Zac Spitzer -
>>  http://zacster.blogspot.com (My Blog)
>>  [phone]
>>  _______________________________________________
>>  mapguide-internals mailing list
>>  mapguide-internals@...
>>  http://lists.osgeo.org/mailman/listinfo/mapguide-internals
>>  _______________________________________________
>>  mapguide-internals mailing list
>>  mapguide-internals@...
>>  http://lists.osgeo.org/mailman/listinfo/mapguide-internals
>>
>
>
>
> --
> Zac Spitzer -
> http://zacster.blogspot.com (My Blog)
> [phone]
>

--
Zac Spitzer -
http://zacster.blogspot.com (My Blog)
[phone]
_______________________________________________
mapguide-internals mailing list
mapguide-internals@...
http://lists.osgeo.org/mailman/listinfo/mapguide-internals