kicad-developers team mailing list archive

[Wayne Stambaugh <stambaughw@xxxxxxxxxxx>]

On 11/12/2014 11:33 AM, Bob Gustafson wrote:
> I tried doing this on one of my blind alleys - I was able to compile
> Boost 1.57 ok,
> but then needed to do it again because the libs need to be static.
> 
>   ./b2 link=static

My guess is that you should be using dynamic libraries in spite of the
bundling issues.  Bernhard, can you give Bob some guidance?

> 
> Then there was the problem of telling kicad where the newly compiled
> boost was..

-DBOOST_ROOT=/path/to/boost-1.57

in the cmake command or

export BOOST_ROOT=/path/to/boost-1.57

before you run cmake.

> 
> ----
> 
> I think a simpler process is to just start from scratch (bzr branch
> lp:kicad)
> 
> and before the 'make' step go into download_boost.cmake and make two
> changes
> - the version from 54 to 57 and on the next line change the MD5 hash result
> (see my notes below)
> 
> From that point, boost will download nicely where kicad expects it to be.
> 
> The problem as you have pointed out is the patch step.
> 
> The first patch works (!) minkowski.hpp
> but then the 2nd patch cstdint.hpp fails

The eventual goal is to do away with our custom patched version of boost
1.54 and use a stock build (except for the gas assembly files required
to build the context library on windows) of boost 1.56 or later and use
find_package() to find boost.  I would also like to move all of the
download_foo.cmake files into a separate kicad dependency library
builder project and only use find_package() when building kicad.

> 
> If there was a way to make a failed patch not kill the build, it would
> be possible to cleanly step over this bump and possibly continue with a
> successful build.
> 
> Bob G
> 
> On 11/12/2014 07:51 AM, Wayne Stambaugh wrote:
>> You should build and install Boost 1.57 outside of kicad by following
>> the boost build instructions of OSX and then configure the kicad build
>> with KICAD_SKIP_BOOST=ON.  You may also have to set BOOST_ROOT if cmake
>> cannot find the installed boost 1.57.  What you are attempting to do
>> will most likely fail because the kicad build configuration applies a
>> bunch of patches to boost 1.54 that will likely not apply cleanly to
>> 1.57.  Sorry if I was not clear about the process.
>>
>> On 11/11/2014 7:58 PM, Bob Gustafson wrote:
>>> To build and test with boost_1_57, my process may be different..
>>>
>>> I went into ~/KiCad/kicad/CMakeModules
>>> and changed in the file download_boost.cmake
>>> the first 54 to 57 - all of the other 54s are in comments
>>>
>>> Then went to ~/KiCad/kicad/kicad and did 'make 2&1 | tee make.out'
>>> The boost 1.57 was downloaded and then failed on the hash check verify
>>> step.
>>>
>>> I then went deeper into download_boost.cmake and changed the check
>>> line to:
>>> set( BOOST_MD5 1be49befbdd9a5ce9def2983ba3e7b76 ) ..
>>>
>>> Then doing the make step over again, it was smart enough not to download
>>> boost again
>>> and it was successful on the verify step, but then it started to patch
>>> and failed.
>>>
>>> I have gone down several blind holes on this, but the build fails every
>>> time.
>>> (Need to go and peel the garlics now)
>>>
>>> ---
>>>
>>> In directory ~/KiCad/kicad/kicad, in file CMakeCache.txt
>>>
>>> Change line 'KICAD_SKIP_BOOST:BOOL=OFF'
>>> to KICAD_SKIP_BOOST:BOOL=ON
>>>
>>> Although it seems that on the first build of Boost,
>>> KICAD_SKIP_BOOST:BOOL should equal OFF (do not skip)
>>> After Boost has been (successfully) built, then the
>>> KICAD_SKIP_BOOST:BOOL can be set to ON (skip..)
>>> I may be wrong about this.
>>>
>>> Later
>>>
>>> Bob G
>>>
>>> On 11/11/2014 12:14 PM, Wayne Stambaugh wrote:
>>>> Thanks Bob.  If the 1.57 testing doesn't cause any issues, I'll commit
>>>> the latest avhttp changes.
>>>>
>>>> Wayne
>>>>
>>>> On 11/11/2014 12:19 PM, Bob Gustafson wrote:
>>>>> Going through your list of instructions for the new avhttp - good
>>>>> instructions by the way.
>>>>>
>>>>> On OSX, boost 1.54 (was on computer) and KICAD_SKIP_BOOST=OFF (this
>>>>> was
>>>>> the setting)
>>>>>
>>>>> Build went fast (no rebuild of boost..) Only a few warnings.
>>>>>
>>>>> kicad executes. Clicking on pcbnew - also looks good. At the top of
>>>>> the
>>>>> pcbnew window, it says: 2014-10-31 BZR 5247.
>>>>>
>>>>> Hope this is what was expected
>>>>>
>>>>> I will build the new boost 1.57 and test that.
>>>>>
>>>>> Bob G
>>>>>
>>>>> On 11/11/2014 08:17 AM, Wayne Stambaugh wrote:
>>>>>> Thank you everyone for testing this.  I verified that it works with
>>>>>> Boost 1.57 using KICAD_SKIP_BOOST=ON and our modified Boost 1.54 with
>>>>>> KICAD_SKIP_BOOST=OFF on windows.  The only unknown at this point is
>>>>>> OSX.
>>>>>>     Would someone please test both Boost build options (stock 1.57
>>>>>> and our
>>>>>> custom 1.54) on OSX?  We have OSX builds in pretty good shape so I
>>>>>> don't
>>>>>> want to break anything.  The process is fairly straight forward.
>>>>>>
>>>>>> 1) Down load the avhttp source zip file from
>>>>>> https://github.com/avplayer/avhttp/archive/master.zip
>>>>>> 2) Copy the zip file to pcbnew/github/avhttp-master.zip.
>>>>>> 3) Delete existing avhttp source folder .downloads-by-cmake/avhttp
>>>>>> 4) Configure and build kicad against boost 1.57 using
>>>>>> KICAD_SKIP_BOOST=ON.
>>>>>> 5) Configure and build kicad against our custom boost 1.54 using
>>>>>> KICAD_SKIP_BOOST=OFF.
>>>>>>
>>>>>> Once I have confirmation that this doesn't break the OSX build, I
>>>>>> will
>>>>>> commit the latest avhttp source zip file.
>>>>>>
>>>>>> Thanks,
>>>>>>
>>>>>> Wayne
>>>>>>
>>>>>> On 11/9/2014 10:45 PM, Blair Bonnett wrote:
>>>>>>> On 10 November 2014 09:41, Nick Østergaard <oe.nick@xxxxxxxxx
>>>>>>> <mailto:oe.nick@xxxxxxxxx>> wrote:
>>>>>>>> Yes, it also builds with kicad's boost and it fetches the
>>>>>>>> footprints
>>>>>>>> form github without problems. So I would say it is safe to update.
>>>>>>>>
>>>>>>> I ran into the same compile error last night after a system update.
>>>>>>> I've
>>>>>>> just compiled with the latest avhttp with both Boost 1.57, and the
>>>>>>> patched 1.54 bundled with KiCad. I encountered no problems using
>>>>>>> either
>>>>>>> -- cvpcb seems happy in both cases.
>>>>>>>
>>>>>>>
>>>>>>> _______________________________________________
>>>>>>> Mailing list: https://launchpad.net/~kicad-developers
>>>>>>> Post to     : kicad-developers@xxxxxxxxxxxxxxxxxxx
>>>>>>> Unsubscribe : https://launchpad.net/~kicad-developers
>>>>>>> More help   : https://help.launchpad.net/ListHelp
>>>>>>>
>>>>>> _______________________________________________
>>>>>> Mailing list: https://launchpad.net/~kicad-developers
>>>>>> Post to     : kicad-developers@xxxxxxxxxxxxxxxxxxx
>>>>>> Unsubscribe : https://launchpad.net/~kicad-developers
>>>>>> More help   : https://help.launchpad.net/ListHelp
>>>>> _______________________________________________
>>>>> Mailing list: https://launchpad.net/~kicad-developers
>>>>> Post to     : kicad-developers@xxxxxxxxxxxxxxxxxxx
>>>>> Unsubscribe : https://launchpad.net/~kicad-developers
>>>>> More help   : https://help.launchpad.net/ListHelp
>>>>>
>>>>
>>>> _______________________________________________
>>>> Mailing list: https://launchpad.net/~kicad-developers
>>>> Post to     : kicad-developers@xxxxxxxxxxxxxxxxxxx
>>>> Unsubscribe : https://launchpad.net/~kicad-developers
>>>> More help   : https://help.launchpad.net/ListHelp
>>>
> 
> 
> _______________________________________________
> Mailing list: https://launchpad.net/~kicad-developers
> Post to     : kicad-developers@xxxxxxxxxxxxxxxxxxx
> Unsubscribe : https://launchpad.net/~kicad-developers
> More help   : https://help.launchpad.net/ListHelp
>