Patch Proposal #1974 Render caching

[Mailing Lists-3]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi Devs

I've been working a little on implementing some render caching to make
QGIS feel more responsive and avoid unnecessary re-rendering of layers
when nothing has really changed.

I've finished writing a simple caching mechanism and would like to ask
you to a) test and b) indicate if you are happy for this to go into trunk.

The patch is here

https://trac.osgeo.org/qgis/attachment/ticket/1974/qgis_rendercachepatch_v7_23Oct2009.diff

Note that caching is disabled by default - you need to enable it in the
Options dialog, rendering tab.


Regards,

- --

Tim Sutton - QGIS Project Steering Committee Member (Release  Manager)
==============================================
Please do not email me off-list with technical
support questions. Using the lists will gain
more exposure for your issues and the knowledge
surrounding your issue will be shared with all.

Visit http://linfiniti.com to find out about:
 * QGIS programming and support services
 * Mapserver and PostGIS based hosting plans
 * FOSS Consulting Services
Skype: timlinux
Irc: timlinux on #qgis at freenode.net
==============================================

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iEYEARECAAYFAkriMRoACgkQqk07qZdiYjetaQCeNJSU2IK/XeCga05gVcEZ5z4c
64EAniqeOlEMReJ7Zv+1ALmugTlmcMMV
=Gz8y
-----END PGP SIGNATURE-----
_______________________________________________
Qgis-developer mailing list
[hidden email]
http://lists.osgeo.org/mailman/listinfo/qgis-developer

[Borys Jurgiel-2]

Dne sobota 24 října 2009 00:41:39 Tim Sutton napsal(a):

> Hi Devs
>
> I've been working a little on implementing some render caching to make
> QGIS feel more responsive and avoid unnecessary re-rendering of layers
> when nothing has really changed.
>
> I've finished writing a simple caching mechanism and would like to ask
> you to a) test and b) indicate if you are happy for this to go into trunk.
>
> The patch is here
>
> https://trac.osgeo.org/qgis/attachment/ticket/1974/qgis_rendercachepatch_v7
> _23Oct2009.diff
>
> Note that caching is disabled by default - you need to enable it in the
> Options dialog, rendering tab.
>
>
> Regards,
>

a) Works fine with rasters (including WMS - now tiling would be great! :D )
with one small issue - no refreshig after layer opacity chage.

b) of course +1
_______________________________________________
Qgis-developer mailing list
[hidden email]
http://lists.osgeo.org/mailman/listinfo/qgis-developer

[Mailing Lists-3]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi
>
> a) Works fine with rasters (including WMS - now tiling would be great! :D )
> with one small issue - no refreshig after layer opacity chage.

Ok this updated patch should fix that:

https://trac.osgeo.org/qgis/attachment/ticket/1974/qgis_rendercachepatch_v824Oct2009.diff

Thanks for testing!

Regards

Tim


>
> b) of course +1
> _______________________________________________
> Qgis-developer mailing list
> [hidden email]
> http://lists.osgeo.org/mailman/listinfo/qgis-developer
>


- --

Tim Sutton - QGIS Project Steering Committee Member (Release  Manager)
==============================================
Please do not email me off-list with technical
support questions. Using the lists will gain
more exposure for your issues and the knowledge
surrounding your issue will be shared with all.

Visit http://linfiniti.com to find out about:
 * QGIS programming and support services
 * Mapserver and PostGIS based hosting plans
 * FOSS Consulting Services
Skype: timlinux
Irc: timlinux on #qgis at freenode.net
==============================================

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iEYEARECAAYFAkriwMkACgkQqk07qZdiYjcbPwCglIJ7yjDxea7jOFgrjIZsgmlD
OioAoMV803bnNGMS5QhA549168RdjHpk
=2EHQ
-----END PGP SIGNATURE-----
_______________________________________________
Qgis-developer mailing list
[hidden email]
http://lists.osgeo.org/mailman/listinfo/qgis-developer

[Borys Jurgiel-2]

Dne sobota 24 října 2009 10:54:40 Tim Sutton napsal(a):

> Ok this updated patch should fix that:
>
> https://trac.osgeo.org/qgis/attachment/ticket/1974/qgis_rendercachepatch_v8
> 24Oct2009.diff

Now is perfect :)
_______________________________________________
Qgis-developer mailing list
[hidden email]
http://lists.osgeo.org/mailman/listinfo/qgis-developer

[Marco Hugentobler]

Hi Tim

Wow, this is really cool and speeds up rendering in many situations.

Now that every layer has its QImage to paint on, it might be a possibility to
render each in its own thread and have great performance on todays quad core
machines.

Regards,
Marco

Am Samstag, 24. Oktober 2009 00.41:39 schrieb Tim Sutton:

> Hi Devs
>
> I've been working a little on implementing some render caching to make
> QGIS feel more responsive and avoid unnecessary re-rendering of layers
> when nothing has really changed.
>
> I've finished writing a simple caching mechanism and would like to ask
> you to a) test and b) indicate if you are happy for this to go into trunk.
>
> The patch is here
>
> https://trac.osgeo.org/qgis/attachment/ticket/1974/qgis_rendercachepatch_v7
>_23Oct2009.diff
>
> Note that caching is disabled by default - you need to enable it in the
> Options dialog, rendering tab.
>
>
> Regards,

--
Dr. Marco Hugentobler
HUGIS - GIS programming and consulting
Honrainweg 5
CH-8038 Zürich
[hidden email]
http://homepage.hispeed.ch/hugis/
_______________________________________________
Qgis-developer mailing list
[hidden email]
http://lists.osgeo.org/mailman/listinfo/qgis-developer

[Andreas Neumann-4]

good - will test this next week.

Marco - do you think the separate QImages could help us with my layer
masking proposal - to be discussed in Vienna or at one of the upcoming
hackfests?

Thanks,
Andreas

Marco Hugentobler wrote:

> Hi Tim
>
> Wow, this is really cool and speeds up rendering in many situations.
>
> Now that every layer has its QImage to paint on, it might be a possibility to
> render each in its own thread and have great performance on todays quad core
> machines.
>
> Regards,
> Marco
>
> Am Samstag, 24. Oktober 2009 00.41:39 schrieb Tim Sutton:
>  
>> Hi Devs
>>
>> I've been working a little on implementing some render caching to make
>> QGIS feel more responsive and avoid unnecessary re-rendering of layers
>> when nothing has really changed.
>>
>> I've finished writing a simple caching mechanism and would like to ask
>> you to a) test and b) indicate if you are happy for this to go into trunk.
>>
>> The patch is here
>>
>> https://trac.osgeo.org/qgis/attachment/ticket/1974/qgis_rendercachepatch_v7
>> _23Oct2009.diff
>>
>> Note that caching is disabled by default - you need to enable it in the
>> Options dialog, rendering tab.
>>
>>
>> Regards,
>>    
>
>  

_______________________________________________
Qgis-developer mailing list
[hidden email]
http://lists.osgeo.org/mailman/listinfo/qgis-developer

[Carson Farmer]

Very nice Tim!

+1 from me, speed is nice, didn't test with rasters though...
Rendering each layer on it's own makes turning them on
and off very responsive, nice work!

Carson

> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Hi Devs
>
> I've been working a little on implementing some render caching to make
> QGIS feel more responsive and avoid unnecessary re-rendering of layers
> when nothing has really changed.
>
> I've finished writing a simple caching mechanism and would like to ask
> you to a) test and b) indicate if you are happy for this to go into trunk.
>
> The patch is here
>
> https://trac.osgeo.org/qgis/attachment/ticket/1974/qgis_rendercachepatch_v7_23Oct2009.diff
>
> Note that caching is disabled by default - you need to enable it in the
> Options dialog, rendering tab.
>
>
> Regards,
>
> - --
>
> Tim Sutton - QGIS Project Steering Committee Member (Release  Manager)
> ==============================================
> Please do not email me off-list with technical
> support questions. Using the lists will gain
> more exposure for your issues and the knowledge
> surrounding your issue will be shared with all.
>
> Visit http://linfiniti.com to find out about:
>  * QGIS programming and support services
>  * Mapserver and PostGIS based hosting plans
>  * FOSS Consulting Services
> Skype: timlinux
> Irc: timlinux on #qgis at freenode.net
> ==============================================
>
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.9 (GNU/Linux)
> Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
>
> iEYEARECAAYFAkriMRoACgkQqk07qZdiYjetaQCeNJSU2IK/XeCga05gVcEZ5z4c
> 64EAniqeOlEMReJ7Zv+1ALmugTlmcMMV
> =Gz8y
> -----END PGP SIGNATURE-----
> _______________________________________________
> Qgis-developer mailing list
> [hidden email]
> http://lists.osgeo.org/mailman/listinfo/qgis-developer
>
>  

--
Carson J. Q. Farmer
ISSP Doctoral Fellow
National Centre for Geocomputation (NCG),
National University of Ireland, Maynooth,
Email: [hidden email]
Web:   http://www.carsonfarmer.com/
       http://www.ftools.ca/

_______________________________________________
Qgis-developer mailing list
[hidden email]
http://lists.osgeo.org/mailman/listinfo/qgis-developer

[Mailing Lists-3]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi

Marco Hugentobler wrote:
> Hi Tim
>
> Wow, this is really cool and speeds up rendering in many situations.
>
> Now that every layer has its QImage to paint on, it might be a possibility to
> render each in its own thread and have great performance on todays quad core
> machines.
>
I remember chatting to Martin about this at the last hackfest. His main
concern was that providers are not thread safe so we could run into
issues e.g. try to run two threads on the same ogr datasource.

I have further plans for the caching system to that can realise speed
improvements when panning and zooming in.

I guess from the feedback there are no objections to me committing this
patch.


Regards

Tim

> Regards,
> Marco
>
> Am Samstag, 24. Oktober 2009 00.41:39 schrieb Tim Sutton:
>> Hi Devs
>>
>> I've been working a little on implementing some render caching to make
>> QGIS feel more responsive and avoid unnecessary re-rendering of layers
>> when nothing has really changed.
>>
>> I've finished writing a simple caching mechanism and would like to ask
>> you to a) test and b) indicate if you are happy for this to go into trunk.
>>
>> The patch is here
>>
>> https://trac.osgeo.org/qgis/attachment/ticket/1974/qgis_rendercachepatch_v7
>> _23Oct2009.diff
>>
>> Note that caching is disabled by default - you need to enable it in the
>> Options dialog, rendering tab.
>>
>>
>> Regards,
>

- --

Tim Sutton - QGIS Project Steering Committee Member (Release  Manager)
==============================================
Please do not email me off-list with technical
support questions. Using the lists will gain
more exposure for your issues and the knowledge
surrounding your issue will be shared with all.

Visit http://linfiniti.com to find out about:
 * QGIS programming and support services
 * Mapserver and PostGIS based hosting plans
 * FOSS Consulting Services
Skype: timlinux
Irc: timlinux on #qgis at freenode.net
==============================================

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iEYEARECAAYFAkri9cIACgkQqk07qZdiYjdFDQCfZaKADQKC8zFEyeEOYcy/GznS
+oQAnioDGKMoOGKMtUVSrNyognonwZ6N
=XW2c
-----END PGP SIGNATURE-----
_______________________________________________
Qgis-developer mailing list
[hidden email]
http://lists.osgeo.org/mailman/listinfo/qgis-developer

[Marco Hugentobler]

Hi Tim

> I remember chatting to Martin about this at the last hackfest. His main
> concern was that providers are not thread safe so we could run into
> issues e.g. try to run two threads on the same ogr datasource.

I think this concern was as we discussed layer tyling and rendering each tile
of a layer in a thread.
If a whole layer renders in a thread, we shouldn't have this problem because
each layer has its own datasource object instance.

Regards,
Marco

Am Samstag, 24. Oktober 2009 14.40:40 schrieb Tim Sutton:

> Hi
>
> Marco Hugentobler wrote:
> > Hi Tim
> >
> > Wow, this is really cool and speeds up rendering in many situations.
> >
> > Now that every layer has its QImage to paint on, it might be a
> > possibility to render each in its own thread and have great performance
> > on todays quad core machines.
>
> I remember chatting to Martin about this at the last hackfest. His main
> concern was that providers are not thread safe so we could run into
> issues e.g. try to run two threads on the same ogr datasource.
>
> I have further plans for the caching system to that can realise speed
> improvements when panning and zooming in.
>
> I guess from the feedback there are no objections to me committing this
> patch.
>
>
> Regards
>
> Tim
>
> > Regards,
> > Marco
> >
> > Am Samstag, 24. Oktober 2009 00.41:39 schrieb Tim Sutton:
> >> Hi Devs
> >>
> >> I've been working a little on implementing some render caching to make
> >> QGIS feel more responsive and avoid unnecessary re-rendering of layers
> >> when nothing has really changed.
> >>
> >> I've finished writing a simple caching mechanism and would like to ask
> >> you to a) test and b) indicate if you are happy for this to go into
> >> trunk.
> >>
> >> The patch is here
> >>
> >> https://trac.osgeo.org/qgis/attachment/ticket/1974/qgis_rendercachepatch
> >>_v7 _23Oct2009.diff
> >>
> >> Note that caching is disabled by default - you need to enable it in the
> >> Options dialog, rendering tab.
> >>
> >>
> >> Regards,

--
Dr. Marco Hugentobler
HUGIS - GIS programming and consulting
Honrainweg 5
CH-8038 Zürich
[hidden email]
http://homepage.hispeed.ch/hugis/
_______________________________________________
Qgis-developer mailing list
[hidden email]
http://lists.osgeo.org/mailman/listinfo/qgis-developer

[Mailing Lists-3]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi

Marco Hugentobler wrote:

> Hi Tim
>
>> I remember chatting to Martin about this at the last hackfest. His main
>> concern was that providers are not thread safe so we could run into
>> issues e.g. try to run two threads on the same ogr datasource.
>
> I think this concern was as we discussed layer tyling and rendering each tile
> of a layer in a thread.
> If a whole layer renders in a thread, we shouldn't have this problem because
> each layer has its own datasource object instance.
>

Ah great! If this is the case it would be something well worth trying
(maybe at the hackfest?).

Regards

Tim

- --

Tim Sutton - QGIS Project Steering Committee Member (Release  Manager)
==============================================
Please do not email me off-list with technical
support questions. Using the lists will gain
more exposure for your issues and the knowledge
surrounding your issue will be shared with all.

Visit http://linfiniti.com to find out about:
 * QGIS programming and support services
 * Mapserver and PostGIS based hosting plans
 * FOSS Consulting Services
Skype: timlinux
Irc: timlinux on #qgis at freenode.net
==============================================

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iEYEARECAAYFAkri/bwACgkQqk07qZdiYje1vQCeIFcqvfH21eBh6VP2yxsrslJM
oOEAoLkjNRYQjTkltMfyW04g7JryboXT
=PTne
-----END PGP SIGNATURE-----
_______________________________________________
Qgis-developer mailing list
[hidden email]
http://lists.osgeo.org/mailman/listinfo/qgis-developer

[Mailing Lists-3]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi all

Ok the patch is applied as of r11831.

Thanks everyone who tested it.

Regards



- --

Tim Sutton - QGIS Project Steering Committee Member (Release  Manager)
==============================================
Please do not email me off-list with technical
support questions. Using the lists will gain
more exposure for your issues and the knowledge
surrounding your issue will be shared with all.

Visit http://linfiniti.com to find out about:
 * QGIS programming and support services
 * Mapserver and PostGIS based hosting plans
 * FOSS Consulting Services
Skype: timlinux
Irc: timlinux on #qgis at freenode.net
==============================================

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iEYEARECAAYFAkrjAf0ACgkQqk07qZdiYje29wCglD3SggFJWOAlWvc9ZD7WoS5A
CsQAn1NlmLLiydW1tVjH+pkkdZnenmw5
=eyU0
-----END PGP SIGNATURE-----
_______________________________________________
Qgis-developer mailing list
[hidden email]
http://lists.osgeo.org/mailman/listinfo/qgis-developer

[Marco Hugentobler]

> Marco - do you think the separate QImages could help us with my layer
> masking proposal - to be discussed in Vienna or at one of the upcoming
> hackfests?

Yes, it is a prerequisite.

One problem with masking on raster level is that everything is rasterized and
so printing to vector svg / pdf will not be possible (or only with image
caching turned off and therefore without masking).

Regards,
Marco

> Am Samstag, 24. Oktober 2009 13.41:07 schrieb Andreas Neumann:
> good - will test this next week.
>
> Marco - do you think the separate QImages could help us with my layer
> masking proposal - to be discussed in Vienna or at one of the upcoming
> hackfests?
>
> Thanks,
> Andreas
>
> Marco Hugentobler wrote:
> > Hi Tim
> >
> > Wow, this is really cool and speeds up rendering in many situations.
> >
> > Now that every layer has its QImage to paint on, it might be a
> > possibility to render each in its own thread and have great performance
> > on todays quad core machines.
> >
> > Regards,
> > Marco
> >
> > Am Samstag, 24. Oktober 2009 00.41:39 schrieb Tim Sutton:
> >> Hi Devs
> >>
> >> I've been working a little on implementing some render caching to make
> >> QGIS feel more responsive and avoid unnecessary re-rendering of layers
> >> when nothing has really changed.
> >>
> >> I've finished writing a simple caching mechanism and would like to ask
> >> you to a) test and b) indicate if you are happy for this to go into
> >> trunk.
> >>
> >> The patch is here
> >>
> >> https://trac.osgeo.org/qgis/attachment/ticket/1974/qgis_rendercachepatch
> >>_v7 _23Oct2009.diff
> >>
> >> Note that caching is disabled by default - you need to enable it in the
> >> Options dialog, rendering tab.
> >>
> >>
> >> Regards,

--
Dr. Marco Hugentobler
HUGIS - GIS programming and consulting
Honrainweg 5
CH-8038 Zürich
[hidden email]
http://homepage.hispeed.ch/hugis/
_______________________________________________
Qgis-developer mailing list
[hidden email]
http://lists.osgeo.org/mailman/listinfo/qgis-developer

[Henri De Feraudy]

The QGis documentation for version 1.0.2 has a section that starts as following
 
16  Creating C++ Applications
Not everyone wants a full blown GIS desktop application. Sometimes you want to just have a widget
inside your application that displays a map while the main goal of the application lies elsewhere. Per-
haps a database frontend with a map display? This Section provides two simple code examples by
Tim Sutton. They are available in the qgis subversion repository together with more interesting tuto-
rials. Check out the whole repository from: https://svn.osgeo.org/qgis/trunk/code_examples/
 
Unfortunately the link provided is not valid.
 
I'm on WIndows XP.
Now I started off hoping I could develop this application just like an ordinary Qt application, using QMake.
Am I deluded? Do I have to learn CMake?
 
All attempts I have tried have generated lots of compiler errors.
Now I have peeked into a file CMakelists.tx that suggests that CORE_EXPORT should be defined as __declspec(dllexport)
and the same definition for GUI_EXPORT
(but this is a guess)
 
Are there any other defines?
On Tim Sutton's blog someone suggested that there be a define
QGISPLUGINDIR=$${QGISPLUGINDIR} CORE_EXPORT= GUI_EXPORT=
This is contradictory with my __declspec(dllexport) (for the last two) and i cant find any occurence of QGISPLUGINDIR
 
Thankyou
Henri
_______________________________________________
Qgis-developer mailing list
[hidden email]
http://lists.osgeo.org/mailman/listinfo/qgis-developer

[Marco Hugentobler]

Hi Henri

Hm, the link https://svn.osgeo.org/qgis/trunk/code_examples/ works for me.

Yes, you may use qmake. Just link your application with libqgis_core.dll and
maybe with libqgis_gui.dll

>CORE_EXPORT= GUI_EXPORT=
is for Linux where you don't need any dllexport macros

>CORE_EXPORT
> should be defined as __declspec(dllexport) and the same definition for
> GUI_EXPORT

Yes, on windows with msvc, you need to define those export macros.

Regards,
Marco

Am Mittwoch, 28. Oktober 2009 15.52:52 schrieb Henri De Feraudy:

> The QGis documentation for version 1.0.2 has a section that starts as
> following
> 16  Creating C++ Applications
> Not everyone wants a full blown GIS desktop application. Sometimes you want
> to just have a widget inside your application that displays a map while the
> main goal of the application lies elsewhere. Per- haps a database frontend
> with a map display? This Section provides two simple code examples by Tim
> Sutton. They are available in the qgis subversion repository together with
> more interesting tuto- rials. Check out the whole repository from:
> https://svn.osgeo.org/qgis/trunk/code_examples/ 
> Unfortunately the link provided is not valid.
>  
> I'm on WIndows XP.
> Now I started off hoping I could develop this application just like an
> ordinary Qt application, using QMake. Am I deluded? Do I have to learn
> CMake?
>  
> All attempts I have tried have generated lots of compiler errors.
> Now I have peeked into a file CMakelists.tx that suggests that CORE_EXPORT
> should be defined as __declspec(dllexport) and the same definition for
> GUI_EXPORT
> (but this is a guess)
>  
> Are there any other defines?
> On Tim Sutton's blog someone suggested that there be a define
> QGISPLUGINDIR=$${QGISPLUGINDIR} CORE_EXPORT= GUI_EXPORT=
> This is contradictory with my __declspec(dllexport) (for the last two) and
> i cant find any occurence of QGISPLUGINDIR
> Thankyou
> Henri
> _______________________________________________
> Qgis-developer mailing list
> [hidden email]
> http://lists.osgeo.org/mailman/listinfo/qgis-developer

--
Dr. Marco Hugentobler
HUGIS - GIS programming and consulting
Webereistrasse 66
CH-8134 Adliswil
[hidden email]
http://homepage.hispeed.ch/hugis
_______________________________________________
Qgis-developer mailing list
[hidden email]
http://lists.osgeo.org/mailman/listinfo/qgis-developer

[Jürgen E. Fischer]

Hi Henri,

On Wed, 28. Oct 2009 at 07:52:52 -0700, Henri De Feraudy wrote:
> This is contradictory with my __declspec(dllexport) (for the last two) and i
> cant find any occurence of QGISPLUGINDIR

__declspec(dllexport) is only useful when building the libraries, because those
export the symbols.  Everywhere else you need __declspec(dllimport).


Jürgen

--
Jürgen E. Fischer         norBIT GmbH               Tel. [phone]
Dipl.-Inf. (FH)           Rheinstraße 13            Fax. [phone]
Software Engineer         D-26506 Norden               http://www.norbit.de

--
norBIT Gesellschaft fuer Unternehmensberatung und Informationssysteme mbH
Rheinstrasse 13, 26506 Norden
GF: Jelto Buurman, HR: Amtsgericht Emden, HRB 5502

_______________________________________________
Qgis-developer mailing list
[hidden email]
http://lists.osgeo.org/mailman/listinfo/qgis-developer