Menu
▾
▴
openfx-developer
|
From: Haarm-Pieter D. <li...@du...> - 2011-02-28 07:10:40
|
Hi, I'm running into an issue where OfxGetNumberOfPlugins isn't being called by Nuke. It's a little bit of an issue because that's where I'd like to insert some licensing code... Nuke appears not to call OfxGetNumberOfPlugins when a file is reopened very quickly after a recent exit or crash (!). Is this expected behavior for an OpenFX host? Can plugin writers plan for any specific ordering of calls to the plugin's public functions? Is there a suggested location to insert licensing calls? In Nuke, the ordering is, sometimes, OfxGetNumberOfPlugins followed by successive calls to OfxGetPlugin. In other cases, OfxGetPlugin is called directly, as if the result of OfxGetNumberOfPlugins was cached between Nuke launches. Either way, for each of call to OfxGetPlugin, Nuke seems to reload the dynamic library entirely as evidenced by repeated instantiations of static variables and calls to library constructors and destructors. I'm testing specifically on OSX with Nuke 5.01v1, 6.1v3 and 6.2v2 for i386 and x86_64. Thanks in advance for any help, HP Duiker Duiker Research |
|
From: Bruno N. <br...@th...> - 2011-03-10 10:46:51
|
Hi Haarm-Pieter, for host specific issues you should contact the host developers directly. For Nuke, try nuk...@su... To avoid continually reloading and re-describing all plugins, Nuke will cache away the information about an OFX plugin when it first encounters it, this is in the NUKE_TMP_DIR/ofxplugincache directory. On Unix-alikes this is /var/tmp/USERID/ofxplugincache So what is probably happening is that Nuke has done the whole load/describe rigmorale the first time round and cached the plugin away. Next time round, as the plugin on disk has not changed, it doesn't load it at all, but it can make menus from the cached descriptions, which means Nuke starts up much faster. So when you create one of your plugins in Nuke, it only then loads the dynamic library containing it, and as it knows which index in the binary that plugin belongs to it is probably skipping the OfxGetNumberOfPlugins call. It still has to call OfxGetPlugin, as otherwise it can't get the needed function pointers for that plugin, also you will be guaranteed that the "On Load" action will be called before any other action is. b On 28 Feb 2011, at 06:40, Haarm-Pieter Duiker wrote: > Hi, > > I'm running into an issue where OfxGetNumberOfPlugins isn't being called by Nuke. It's a little bit of an issue because that's where I'd like to insert some licensing code... Nuke appears not to call OfxGetNumberOfPlugins when a file is reopened very quickly after a recent exit or crash (!). Is this expected behavior for an OpenFX host? Can plugin writers plan for any specific ordering of calls to the plugin's public functions? Is there a suggested location to insert licensing calls? > > In Nuke, the ordering is, sometimes, OfxGetNumberOfPlugins followed by successive calls to OfxGetPlugin. In other cases, OfxGetPlugin is called directly, as if the result of OfxGetNumberOfPlugins was cached between Nuke launches. Either way, for each of call to OfxGetPlugin, Nuke seems to reload the dynamic library entirely as evidenced by repeated instantiations of static variables and calls to library constructors and destructors. > > I'm testing specifically on OSX with Nuke 5.01v1, 6.1v3 and 6.2v2 for i386 and x86_64. > > Thanks in advance for any help, > HP Duiker > Duiker Research > ------------------------------------------------------------------------------ > Free Software Download: Index, Search & Analyze Logs and other IT data in > Real-Time with Splunk. Collect, index and harness all the fast moving IT data > generated by your applications, servers and devices whether physical, virtual > or in the cloud. Deliver compliance at lower cost and gain new business > insights. http://p.sf.net/sfu/splunk-dev2dev _______________________________________________ > Openfx-developer mailing list > Ope...@li... > https://lists.sourceforge.net/lists/listinfo/openfx-developer -- Bruno Nicoletti, Chief Technology Officer The Foundry 6th Floor, The Communications Building, 48, Leicester Square, London, WC2H 7LT, UK Skype: brunonicoletti Tel: +44 (0)20 7968 6828 - Fax: +44 (0)20 7930 8906 Web: www.thefoundry.co.uk Email: br...@th... The Foundry Visionmongers Ltd. Registered in England and Wales No: 4642027 |
|
From: Haarm-Pieter D. <li...@du...> - 2011-03-15 21:00:49
|
Hi Bruno, Thanks for the help. I was able to work through this issue to some degree. It's not ideal and other hosts don't take the same approach of loading the same .ofx library multiple times, but the affect now seems to be harmless. Regards, HP Duiker Duiker Research On Thu, Mar 10, 2011 at 2:46 AM, Bruno Nicoletti <br...@th...>wrote: > Hi Haarm-Pieter, > > for host specific issues you should contact the host developers directly. > For Nuke, try nuk...@su... > > To avoid continually reloading and re-describing all plugins, Nuke will > cache away the information about an OFX plugin when it first encounters it, > this is in the NUKE_TMP_DIR/ofxplugincache directory. On Unix-alikes this is > /var/tmp/USERID/ofxplugincache > > So what is probably happening is that Nuke has done the whole load/describe > rigmorale the first time round and cached the plugin away. Next time round, > as the plugin on disk has not changed, it doesn't load it at all, but it can > make menus from the cached descriptions, which means Nuke starts up much > faster. So when you create one of your plugins in Nuke, it only then loads > the dynamic library containing it, and as it knows which index in the binary > that plugin belongs to it is probably skipping the OfxGetNumberOfPlugins > call. > > It still has to call OfxGetPlugin, as otherwise it can't get the needed > function pointers for that plugin, also you will be guaranteed that the "On > Load" action will be called before any other action is. > > b > > > On 28 Feb 2011, at 06:40, Haarm-Pieter Duiker wrote: > > > Hi, > > > > I'm running into an issue where OfxGetNumberOfPlugins isn't being called > by Nuke. It's a little bit of an issue because that's where I'd like to > insert some licensing code... Nuke appears not to call OfxGetNumberOfPlugins > when a file is reopened very quickly after a recent exit or crash (!). Is > this expected behavior for an OpenFX host? Can plugin writers plan for any > specific ordering of calls to the plugin's public functions? Is there a > suggested location to insert licensing calls? > > > > In Nuke, the ordering is, sometimes, OfxGetNumberOfPlugins followed by > successive calls to OfxGetPlugin. In other cases, OfxGetPlugin is called > directly, as if the result of OfxGetNumberOfPlugins was cached between Nuke > launches. Either way, for each of call to OfxGetPlugin, Nuke seems to reload > the dynamic library entirely as evidenced by repeated instantiations of > static variables and calls to library constructors and destructors. > > > > I'm testing specifically on OSX with Nuke 5.01v1, 6.1v3 and 6.2v2 for > i386 and x86_64. > > > > Thanks in advance for any help, > > HP Duiker > > Duiker Research > > > ------------------------------------------------------------------------------ > > Free Software Download: Index, Search & Analyze Logs and other IT data in > > Real-Time with Splunk. Collect, index and harness all the fast moving IT > data > > generated by your applications, servers and devices whether physical, > virtual > > or in the cloud. Deliver compliance at lower cost and gain new business > > insights. http://p.sf.net/sfu/splunk-dev2dev_______________________________________________ > > Openfx-developer mailing list > > Ope...@li... > > https://lists.sourceforge.net/lists/listinfo/openfx-developer > > -- > Bruno Nicoletti, > Chief Technology Officer > The Foundry > 6th Floor, The Communications Building, > 48, Leicester Square, > London, WC2H 7LT, UK > Skype: brunonicoletti > Tel: +44 (0)20 7968 6828 - Fax: +44 (0)20 7930 8906 > Web: www.thefoundry.co.uk > Email: br...@th... > > The Foundry Visionmongers Ltd. > Registered in England and Wales No: 4642027 > > > > ------------------------------------------------------------------------------ > Colocation vs. Managed Hosting > A question and answer guide to determining the best fit > for your organization - today and in the future. > http://p.sf.net/sfu/internap-sfd2d > _______________________________________________ > Openfx-developer mailing list > Ope...@li... > https://lists.sourceforge.net/lists/listinfo/openfx-developer > |
|
From: Bruno N. <br...@th...> - 2011-03-16 13:08:57
|
Happy it was resolved. b On 15 Mar 2011, at 20:38, Haarm-Pieter Duiker wrote: > Hi Bruno, > > Thanks for the help. I was able to work through this issue to some degree. It's not ideal and other hosts don't take the same approach of loading the same .ofx library multiple times, but the affect now seems to be harmless. > > Regards, > HP Duiker > Duiker Research > > On Thu, Mar 10, 2011 at 2:46 AM, Bruno Nicoletti <br...@th...> wrote: > Hi Haarm-Pieter, > > for host specific issues you should contact the host developers directly. For Nuke, try nuk...@su... > > To avoid continually reloading and re-describing all plugins, Nuke will cache away the information about an OFX plugin when it first encounters it, this is in the NUKE_TMP_DIR/ofxplugincache directory. On Unix-alikes this is /var/tmp/USERID/ofxplugincache > > So what is probably happening is that Nuke has done the whole load/describe rigmorale the first time round and cached the plugin away. Next time round, as the plugin on disk has not changed, it doesn't load it at all, but it can make menus from the cached descriptions, which means Nuke starts up much faster. So when you create one of your plugins in Nuke, it only then loads the dynamic library containing it, and as it knows which index in the binary that plugin belongs to it is probably skipping the OfxGetNumberOfPlugins call. > > It still has to call OfxGetPlugin, as otherwise it can't get the needed function pointers for that plugin, also you will be guaranteed that the "On Load" action will be called before any other action is. > > b > > > On 28 Feb 2011, at 06:40, Haarm-Pieter Duiker wrote: > > > Hi, > > > > I'm running into an issue where OfxGetNumberOfPlugins isn't being called by Nuke. It's a little bit of an issue because that's where I'd like to insert some licensing code... Nuke appears not to call OfxGetNumberOfPlugins when a file is reopened very quickly after a recent exit or crash (!). Is this expected behavior for an OpenFX host? Can plugin writers plan for any specific ordering of calls to the plugin's public functions? Is there a suggested location to insert licensing calls? > > > > In Nuke, the ordering is, sometimes, OfxGetNumberOfPlugins followed by successive calls to OfxGetPlugin. In other cases, OfxGetPlugin is called directly, as if the result of OfxGetNumberOfPlugins was cached between Nuke launches. Either way, for each of call to OfxGetPlugin, Nuke seems to reload the dynamic library entirely as evidenced by repeated instantiations of static variables and calls to library constructors and destructors. > > > > I'm testing specifically on OSX with Nuke 5.01v1, 6.1v3 and 6.2v2 for i386 and x86_64. > > > > Thanks in advance for any help, > > HP Duiker > > Duiker Research > > ------------------------------------------------------------------------------ > > Free Software Download: Index, Search & Analyze Logs and other IT data in > > Real-Time with Splunk. Collect, index and harness all the fast moving IT data > > generated by your applications, servers and devices whether physical, virtual > > or in the cloud. Deliver compliance at lower cost and gain new business > > insights. http://p.sf.net/sfu/splunk-dev2dev _______________________________________________ > > Openfx-developer mailing list > > Ope...@li... > > https://lists.sourceforge.net/lists/listinfo/openfx-developer > > -- > Bruno Nicoletti, > Chief Technology Officer > The Foundry > 6th Floor, The Communications Building, > 48, Leicester Square, > London, WC2H 7LT, UK > Skype: brunonicoletti > Tel: +44 (0)20 7968 6828 - Fax: +44 (0)20 7930 8906 > Web: www.thefoundry.co.uk > Email: br...@th... > > The Foundry Visionmongers Ltd. > Registered in England and Wales No: 4642027 > > > ------------------------------------------------------------------------------ > Colocation vs. Managed Hosting > A question and answer guide to determining the best fit > for your organization - today and in the future. > http://p.sf.net/sfu/internap-sfd2d > _______________________________________________ > Openfx-developer mailing list > Ope...@li... > https://lists.sourceforge.net/lists/listinfo/openfx-developer > > ------------------------------------------------------------------------------ > Colocation vs. Managed Hosting > A question and answer guide to determining the best fit > for your organization - today and in the future. > http://p.sf.net/sfu/internap-sfd2d_______________________________________________ > Openfx-developer mailing list > Ope...@li... > https://lists.sourceforge.net/lists/listinfo/openfx-developer -- Bruno Nicoletti, Chief Technology Officer The Foundry 6th Floor, The Communications Building, 48, Leicester Square, London, WC2H 7LT, UK Skype: brunonicoletti Tel: +44 (0)20 7968 6828 - Fax: +44 (0)20 7930 8906 Web: www.thefoundry.co.uk Email: br...@th... The Foundry Visionmongers Ltd. Registered in England and Wales No: 4642027 |
|
From: Mike W. <sm...@ci...> - 2011-04-20 04:21:46
|
Hi there, I come across this 'analysis pass' term in the 1.2 documentation but I am not sure what it is referring to exactly as there is not much mentioned in the whole doc. Does it mean one may set this flag "kOfxImageEffectPropInAnalysis" during rendering and what exactly does it serve? Thanks. Mike |
|
From: Pierre J. <ja...@re...> - 2011-04-21 05:37:04
|
On 4/19/2011 9:05 PM, Mike Wong wrote: > Hi there, > > I come across this 'analysis pass' term in the 1.2 documentation but I am not sure what it is referring to exactly as there is not much mentioned in the whole doc. Does it mean one may set this flag "kOfxImageEffectPropInAnalysis" during rendering and what exactly does it serve? The theory is if you need to write parameter values based on analyzing the image content, some applications don't like you do that at the actual render stage. > Thanks. > > Mike > > ------------------------------------------------------------------------------ > Benefiting from Server Virtualization: Beyond Initial Workload > Consolidation -- Increasing the use of server virtualization is a top > priority.Virtualization can reduce costs, simplify management, and improve > application availability and disaster protection. Learn more about boosting > the value of server virtualization. http://p.sf.net/sfu/vmware-sfdev2dev > _______________________________________________ > Openfx-developer mailing list > Ope...@li... > https://lists.sourceforge.net/lists/listinfo/openfx-developer > |
|
From: Mike W. <sm...@ci...> - 2011-04-21 06:08:57
|
Pierre, Thanks for the info. As there's no specific callback for this so-called 'analysis' pass, Is it up to the developer when to set "kOfxImageEffectPropInAnalysis" flag as 1 and then carry out computation during for example e.g. kOfxActionInstanceChanged callback? > > Hi there, > > > > I come across this 'analysis pass' term in the 1.2 documentation but > I am not sure what it is referring to exactly as there is not much > mentioned in the whole doc. Does it mean one may set this flag > "kOfxImageEffectPropInAnalysis" during rendering and what exactly does > it serve? > > The theory is if you need to write parameter values based on analyzing > the image content, some applications don't like you do that > at the actual render stage. |
|
From: Pierre J. <ja...@re...> - 2011-04-21 07:38:49
|
On 4/20/2011 11:08 PM, Mike Wong wrote: > Pierre, > > Thanks for the info. > > As there's no specific callback for this so-called 'analysis' pass, Is it up to the developer when to set "kOfxImageEffectPropInAnalysis" flag as 1 and then carry out computation during for example e.g. kOfxActionInstanceChanged callback? Actually I was working from memory instead of what the header files actually say and I see it says that This feature is likely to be deprecated in future releases. So I am not clear. Right now here we only set param values during interact callback ( like in overlay.cpp example - gPropHost->propSetPointer(effectProps, kOfxImageEffectPluginPropOverlayInteractV1, 0, (void *) overlayMain); in describe action ). You can also maybe maintain instance data. || >>> Hi there, >>> >>> I come across this 'analysis pass' term in the 1.2 documentation but >> I am not sure what it is referring to exactly as there is not much >> mentioned in the whole doc. Does it mean one may set this flag >> "kOfxImageEffectPropInAnalysis" during rendering and what exactly does >> it serve? >> >> The theory is if you need to write parameter values based on analyzing >> the image content, some applications don't like you do that >> at the actual render stage. > > > > > ------------------------------------------------------------------------------ > Benefiting from Server Virtualization: Beyond Initial Workload > Consolidation -- Increasing the use of server virtualization is a top > priority.Virtualization can reduce costs, simplify management, and improve > application availability and disaster protection. Learn more about boosting > the value of server virtualization. http://p.sf.net/sfu/vmware-sfdev2dev > _______________________________________________ > Openfx-developer mailing list > Ope...@li... > https://lists.sourceforge.net/lists/listinfo/openfx-developer > |
|
From: Bruno N. <br...@th...> - 2011-04-21 08:36:22
|
Exactly. Setting parameters during a render is a 'Bad Thing', as on some hosts it will either simply not work, or it will invalidate the output you just computed and cause a re-render, which will invalidate the output you computed, which will re-render and so on. b On 21 Apr 2011, at 07:08, Mike Wong wrote: > > Pierre, > > Thanks for the info. > > As there's no specific callback for this so-called 'analysis' pass, Is it up to the developer when to set "kOfxImageEffectPropInAnalysis" flag as 1 and then carry out computation during for example e.g. kOfxActionInstanceChanged callback? > >>> Hi there, >>> >>> I come across this 'analysis pass' term in the 1.2 documentation but >> I am not sure what it is referring to exactly as there is not much >> mentioned in the whole doc. Does it mean one may set this flag >> "kOfxImageEffectPropInAnalysis" during rendering and what exactly does >> it serve? >> >> The theory is if you need to write parameter values based on analyzing >> the image content, some applications don't like you do that >> at the actual render stage. > > > > > > ------------------------------------------------------------------------------ > Benefiting from Server Virtualization: Beyond Initial Workload > Consolidation -- Increasing the use of server virtualization is a top > priority.Virtualization can reduce costs, simplify management, and improve > application availability and disaster protection. Learn more about boosting > the value of server virtualization. http://p.sf.net/sfu/vmware-sfdev2dev > _______________________________________________ > Openfx-developer mailing list > Ope...@li... > https://lists.sourceforge.net/lists/listinfo/openfx-developer -- Bruno Nicoletti, Chief Technology Officer The Foundry 6th Floor, The Communications Building, 48, Leicester Square, London, WC2H 7LT, UK Skype: brunonicoletti Tel: +44 (0)20 7968 6828 - Fax: +44 (0)20 7930 8906 Web: www.thefoundry.co.uk Email: br...@th... The Foundry Visionmongers Ltd. Registered in England and Wales No: 4642027 |
|
From: Fabien C. <fc...@qu...> - 2011-04-21 11:14:01
|
Bruno Nicoletti a écrit : > Exactly. > > Setting parameters during a render is a 'Bad Thing', as on some hosts it will either simply not work, or it will invalidate the output you just computed and cause a re-render, which will invalidate the output you computed, which will re-render and so on. > As it's written in the documentation that the property kOfxImageEffectPropInAnalysis "is likely to be deprecated in future releases", what the new solution to do this? Or maybe as it's only a reaction to an instance changed action, the host don't really need to know it? So we need nothing to set ? In practice this means that we need to have a button to update the internal values ? In the particular case, we want to have our analysis pass (frame independent) completly transparent to users (so we recompute the values first at each new render), but we want these values to be visibles to users (into disabled parameters). Is it a bad thing to modify, within the render, parameters declared with the property kOfxParamPropEvaluateOnChange = 0 ? Fabien |
|
From: Pierre J. <ja...@re...> - 2011-04-21 16:33:34
|
On 4/21/2011 12:24 AM, Pierre Jasmin wrote: > On 4/20/2011 11:08 PM, Mike Wong wrote: >> Pierre, >> >> Thanks for the info. >> >> As there's no specific callback for this so-called 'analysis' pass, Is it up to the developer when to set "kOfxImageEffectPropInAnalysis" flag as 1 and then carry out computation during for example e.g. kOfxActionInstanceChanged callback? > > Actually I was working from memory instead of what the header files actually say and I see it says > that This feature is likely to be deprecated in future releases. So I am not clear. > > Right now here we only set param values during interact callback ( like in overlay.cpp example - > gPropHost->propSetPointer(effectProps, kOfxImageEffectPluginPropOverlayInteractV1, 0, (void *) overlayMain); in describe action > ). You can also maybe maintain instance data. I lied we of course also set params in actionInstanceChanged all over the place (particularly for hosts that don't support TypeBoolean and TypeChoice animation) I am also seeing kOfxParamPropPluginMayWrite I never noticed before, does not seem any host force us to tag that to each parameter that potentially can paramSetValue. Shouldn't that also be "likely to be deprecated in future releases" or there is something I am missing here? > > > > || >>>> Hi there, >>>> >>>> I come across this 'analysis pass' term in the 1.2 documentation but >>> I am not sure what it is referring to exactly as there is not much >>> mentioned in the whole doc. Does it mean one may set this flag >>> "kOfxImageEffectPropInAnalysis" during rendering and what exactly does >>> it serve? >>> >>> The theory is if you need to write parameter values based on analyzing >>> the image content, some applications don't like you do that >>> at the actual render stage. >> >> >> >> ------------------------------------------------------------------------------ >> Benefiting from Server Virtualization: Beyond Initial Workload >> Consolidation -- Increasing the use of server virtualization is a top >> priority.Virtualization can reduce costs, simplify management, and improve >> application availability and disaster protection. Learn more about boosting >> the value of server virtualization.http://p.sf.net/sfu/vmware-sfdev2dev >> _______________________________________________ >> Openfx-developer mailing list >> Ope...@li... >> https://lists.sourceforge.net/lists/listinfo/openfx-developer >> > > > ------------------------------------------------------------------------------ > Benefiting from Server Virtualization: Beyond Initial Workload > Consolidation -- Increasing the use of server virtualization is a top > priority.Virtualization can reduce costs, simplify management, and improve > application availability and disaster protection. Learn more about boosting > the value of server virtualization. http://p.sf.net/sfu/vmware-sfdev2dev > > > _______________________________________________ > Openfx-developer mailing list > Ope...@li... > https://lists.sourceforge.net/lists/listinfo/openfx-developer |
|
From: Bruno N. <br...@th...> - 2011-04-21 16:36:44
|
On 21 Apr 2011, at 17:32, Pierre Jasmin wrote: > On 4/21/2011 12:24 AM, Pierre Jasmin wrote: >> On 4/20/2011 11:08 PM, Mike Wong wrote: >>> Pierre, >>> >>> Thanks for the info. >>> >>> As there's no specific callback for this so-called 'analysis' pass, Is it up to the developer when to set "kOfxImageEffectPropInAnalysis" flag as 1 and then carry out computation during for example e.g. kOfxActionInstanceChanged callback? >>> >> >> Actually I was working from memory instead of what the header files actually say and I see it says >> that This feature is likely to be deprecated in future releases. So I am not clear. >> >> Right now here we only set param values during interact callback ( like in overlay.cpp example - gPropHost->propSetPointer(effectProps, kOfxImageEffectPluginPropOverlayInteractV1, 0, (void *) overlayMain); in describe action ). You can also maybe maintain instance data. > > I lied we of course also set params in actionInstanceChanged all over the place (particularly for hosts that don't support TypeBoolean and TypeChoice animation) > > I am also seeing kOfxParamPropPluginMayWrite I never noticed before, does not seem any host force us to tag that to each parameter that potentially can paramSetValue. > Shouldn't that also be "likely to be deprecated in future releases" or there is something I am missing here? The theory was that hosts might appreciate that property as a hint, however in practice it seems to be redundant. -- Bruno Nicoletti, Chief Technology Officer The Foundry 6th Floor, The Communications Building, 48, Leicester Square, London, WC2H 7LT, UK Skype: brunonicoletti Tel: +44 (0)20 7968 6828 - Fax: +44 (0)20 7930 8906 Web: www.thefoundry.co.uk Email: br...@th... The Foundry Visionmongers Ltd. Registered in England and Wales No: 4642027 |
|
From: Pierre J. <ja...@re...> - 2011-04-25 21:15:22
|
Peter, Hope you are doing good, wish your company was an active member so we could pick your big brain :) Not trying to single you out here, but I vaguely remember back then you had an objection about when it's kosher to write params... (and my active mail box does not extend that far in the past) so I am thinking you probably have a strong opinion on this: Before we deprecate any flags (or adapt their meaning) it would probably be useful to have a best practice documentation note. I understand that some hosts might have some issues with param writing during the render call, so here's a simple scenario: A user presses a push button (or equivalent if not supported) and as a result of that event, the plugin wants to scan all the frames in the sequence (get and release images), write at each frame something in some parameters (e.g. the min and max values of each frame to keep the example simple). a. In which action is that allowed in your apps? b. Is it useful for you to know or not ahead that the plugin wants to do that ? If so when? c. Any other comments on that simple example? Mine is: Theorically there is sometimes an alternative which would be to store that in instance data and drop that in some custom param at OfxActionSyncPrivateData but that is also proving not reliable right now in many hosts. I am thinking if there was a clear documented note for that then maybe what|kOfxImageEffectPropInAnalysis| means would have a reference, and then at least it would have a reason to live (e.g.: to allow a plugin to bail out during initialization). Pierre On 4/21/2011 9:36 AM, Bruno Nicoletti wrote: > On 21 Apr 2011, at 17:32, Pierre Jasmin wrote: > >> On 4/21/2011 12:24 AM, Pierre Jasmin wrote: >>> On 4/20/2011 11:08 PM, Mike Wong wrote: >>>> Pierre, >>>> >>>> Thanks for the info. >>>> >>>> As there's no specific callback for this so-called 'analysis' pass, Is it up to the developer when to set "kOfxImageEffectPropInAnalysis" flag as 1 and then carry out computation during for example e.g. kOfxActionInstanceChanged callback? >>>> >>> Actually I was working from memory instead of what the header files actually say and I see it says >>> that This feature is likely to be deprecated in future releases. So I am not clear. >>> >>> Right now here we only set param values during interact callback ( like in overlay.cpp example - gPropHost->propSetPointer(effectProps, kOfxImageEffectPluginPropOverlayInteractV1, 0, (void *) overlayMain); in describe action ). You can also maybe maintain instance data. >> I lied we of course also set params in actionInstanceChanged all over the place (particularly for hosts that don't support TypeBoolean and TypeChoice animation) >> >> I am also seeing kOfxParamPropPluginMayWrite I never noticed before, does not seem any host force us to tag that to each parameter that potentially can paramSetValue. >> Shouldn't that also be "likely to be deprecated in future releases" or there is something I am missing here? > The theory was that hosts might appreciate that property as a hint, however in practice it seems to be redundant. > > > -- > Bruno Nicoletti, > Chief Technology Officer > The Foundry > 6th Floor, The Communications Building, > 48, Leicester Square, > London, WC2H 7LT, UK > Skype: brunonicoletti > Tel: +44 (0)20 7968 6828 - Fax: +44 (0)20 7930 8906 > Web: www.thefoundry.co.uk > Email: br...@th... > > The Foundry Visionmongers Ltd. > Registered in England and Wales No: 4642027 > > |
×
Want the latest updates on software, tech news, and AI?
Get latest updates about software, tech news, and AI from SourceForge directly in your inbox once a month.
Thanks for helping keep SourceForge clean.
X