> On Feb. 8, 2011, 10:41 a.m., Boroondas Gupte wrote:
> > > If there is some better way to more exactly target these two items please 
> > > point it out.
> > 
> > You should be able to get the same effect when wrapping the only place 
> > where this file is referenced in a LL_TESTS condition, i.e., change 
> > https://bitbucket.org/lindenlab/viewer-development/src/b0bceb572090/indra/CMakeLists.txt#cl-74
> >  from
> > if (NOT LINUX)
> > to
> > if (LL_TESTS AND NOT LINUX)
> > (and the same for the endif, of course)
> > 
> > Whether that's a better place, I don't know.
> > 
> > Though, I think LL_TESTS is the wrong conditional here, anyway. LL_TESTS is 
> > for enabling unit and integration tests. llplugintest however, I have 
> > learned on IRC today, seems to be a fully separate program based on and 
> > similar to uBrowser that could be used to load and test individual 
> > llmediaplugins, would it communicate with them in the same way the viewer 
> > does. (Which, according to MichelleZ, it doesn't, thus potentially 
> > misleading developers of new plugins.)
> > 
> > It should probably not be built unless explicitly requested, thus a new 
> > variable, defaulting to OFF and different from LL_TESTS would suit this 
> > much better. Or, just delete the referencing at 
> > https://bitbucket.org/lindenlab/viewer-development/src/b0bceb572090/indra/CMakeLists.txt#cl-75
> >  completely, thus unconditionally removing it from the dependencies of the 
> > viewer itself, and have those that want to build it explicitly state it as 
> > a build target.
> 
> Merov Linden wrote:
>     I'm afraid that, if that app is not built by default, it will rapidly rot 
> and become unmaintainable. There is value having it built (even if not used) 
> by LL devs and TC at least (who all build with LL_TESTS on): make sure we 
> don't add viewer dependencies where we don't need any and maintain a clean 
> separation of concerns between plugins and hosts.

I can see that point. Though, I'd still like a separate variable, if alone to 
avoid people from mistaking llplugintest for a unit or integration test, like I 
initially did. Just have that (new) variable default on ON, then, rather than 
OFF.


- Boroondas


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
http://codereview.secondlife.com/r/144/#review347
-----------------------------------------------------------


On Feb. 16, 2011, 11:32 a.m., Jonathan Yap wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> http://codereview.secondlife.com/r/144/
> -----------------------------------------------------------
> 
> (Updated Feb. 16, 2011, 11:32 a.m.)
> 
> 
> Review request for Viewer.
> 
> 
> Summary
> -------
> 
> llmediaplugintest shows up to be compiled even though -DLL_TESTS:BOOL=OFF has 
> been used on the command line.
> 
> This cmake file does not use the call to LL_ADD_PROJECT_UNIT_TESTS that other 
> cmake files do.  
> 
> LL_ADD_PROJECT_UNIT_TESTS is usually wrapped with if(LL_TESTS)
> 
> I could not figure out which lines suppress the inclusion of 
> copy_plugintest_libs and llmediaplugintest into the list of what is to be 
> built so wrapped the entire file around if(LL_TESTS).  
> 
> If there is some better way to more exactly target these two items please 
> point it out.
> 
> 
> This addresses bug STORM-977.
>     http://jira.secondlife.com/browse/STORM-977
> 
> 
> Diffs
> -----
> 
>   indra/CMakeLists.txt b0bceb572090 
>   indra/test_apps/llplugintest/CMakeLists.txt b0bceb572090 
> 
> Diff: http://codereview.secondlife.com/r/144/diff
> 
> 
> Testing
> -------
> 
> 
> Thanks,
> 
> Jonathan
> 
>

_______________________________________________
Policies and (un)subscribe information available here:
http://wiki.secondlife.com/wiki/OpenSource-Dev
Please read the policies before posting to keep unmoderated posting privileges

Reply via email to