Reinplace

[Craig Treleaven]

At 1:10 AM -0500 4/3/12, Ryan Schmidt wrote:
>On Apr 2, 2012, at 19:55, Craig Treleaven wrote:
>
>>  At 4:27 PM -0500 4/2/12, Ryan Schmidt wrote:
>>>  On Apr 2, 2012, at 14:35, Joshua Root wrote:
>>>
>>>>  Does just "find ${configure.dir} -name Makefile" also give unexpected
>>>>  results?
>>>
>>>  The "find" correctly finds 28 files named "Makefile". It 
>>>correctly removes " -L " from 22 of them. The remaining 6 are in 
>>>the directory mythplugins/mythweather/mythweather/scripts and do 
>>>not contain that string so there's nothing to remove.
>>
>>  Ryan:
>>
>>  There should be 58 Makefiles, see attached list.
>>
>>  Craig
>>  <MythTV_Makefile_list.txt>
>
>On my system, after running configure, there are 28 files named 
>"Makefile". The list you attached shows 46 files called "Makefile", 
>so I guess 18 of them were created after the configure phase, which 
>means they're going to be rather hard to patch. This patching is a 
>workaround anyway. We should be figuring out what's writing " -L " 
>into the Makefiles (something in the configure script) and changing 
>that. Or, asking the mythtv developers to fix it, since it's their 
>software.

I think you've hit the nail on the head--the problematic Makefiles 
have a later time stamp.  I didn't know it was possible that they 
could be recreated during the build phase.

Obviously I agree that fixing the generation of the spurious -L's in 
the first place would be preferable.  I spent an inordinate amount of 
time last week trying to track down the source but I don't know qmake 
(or make, to be honest) and had no success.

Come to think of it, I focused only on the configure stage--comparing 
the environment variables set and the arguments passed to 
./configure.  If these spurious -L's are happening at the _build_ 
stage, I need to go through those.  I'll give that a go this morning.

Craig