discuss-gnuradio
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Discuss-gnuradio] Toggle use of bad versions of Boost


From: Tom Rondeau
Subject: Re: [Discuss-gnuradio] Toggle use of bad versions of Boost
Date: Tue, 26 Feb 2013 11:21:28 -0500

On Tue, Feb 26, 2013 at 11:10 AM, Alexandru Csete <address@hidden> wrote:
> On Tue, Feb 26, 2013 at 3:15 PM, Tom Rondeau <address@hidden> wrote:
>> On Tue, Feb 26, 2013 at 7:01 AM, Alexandru Csete <address@hidden> wrote:
>>> On Wed, Feb 13, 2013 at 8:12 PM, Tom Rondeau <address@hidden> wrote:
>>>> GNU Radio: now allowing you to shoot yourself in the foot!
>>>>
>>>> We've identified a few versions of Boost that don't work well with GNU
>>>> Radio, 1.46, 1.47, and 1.52. A little while ago, I updated our cmake build
>>>> files to look for these specific versions and disable their use if found.
>>>>
>>>> Sadly, many distros currently ship with one of these known bad versions.
>>>> Looking at Boost's website, 1.47 and 1.52 are by far the most downloaded
>>>> Boost versions. So, as recommended by Johnathan, I put in a cmake command
>>>> line option that allows you to override the version check so GNU Radio will
>>>> build with any current Boost.
>>>>
>>>> To use it, just specify "-DENABLE_BAD_BOOST=True" on your cmake command
>>>> line. By default this is False. That will allow you to build against
>>>> anything. If this value is False, all GNU Radio components will be disabled
>>>> if it finds one of the bad versions. (VOLK and a couple of others will 
>>>> still
>>>> pass and compile since they are not affected by the problem).
>>>>
>>>> Just be aware that using one of these versions can cause problems.
>>>> Specifically, it can lock up your application when shutting down a
>>>> flowgraph. Also, many distros that ship with one of these bad versions also
>>>> have other, good, versions that you can pretty easily switch to.
>>>>
>>>
>>> Hi,
>>>
>>> Sorry for the late reply, but I just realized that half of my
>>> computers running Mint 13 (based on Ubuntu 12.04) come with Boost
>>> 1.46, yet I have had no problem building gnuradio without using
>>> "-DENABLE_BAD_BOOST=True"
>>> Admittedly, this distro appears to have a horrible mix of 1.46 and
>>> 1.48, which may confuse cmake - I really have no idea which one really
>>> is installed but I do have runtime issues on these computers. If
>>> anybody has ideas for what we can do on Mint 13 / Ubuntu 12.04, please
>>> let me know.
>>>
>>> Alex
>>
>> Alex,
>>
>> You can see which Boost version Cmake is finding for you by looking at
>> CMakeCache.txt in the top of the build directory. Check out the
>> "Boost_<library>_LIBRARY" and "Boost_INCLUDE_DIR" to see which ones
>> were found and are being used. Hopefully, it's finding the 1.48
>> version that's installed.
>
> Actually, it seems that I only have version 1.46 libraries installed,
> see the attached file.
>
> This is really screwed up: I have libboost-dev installed, which
> according to synaptic is at version 1.48.0.2. Looking in the package
> info I see that it depends on libboost1.46-dev, which explain why all
> my boost libs are version 1.46. The same applies to many other boost
> packages, e.g. libboost-thread-dev.

Grep CMakeCache.txt for "Boost_VERSION" and see what that returns. For
1.48, it'll look like 104800, and for 1.46 it should be 104600.

Also see what it tells you when you grep for ENABLE_BAD_BOOST. If this
is set to True, it will allow any version.

> Cmake didn't complain about bad boost version and I can see no version
> info in CMakeCache.txt. Gnuradio compiles and install just fine. I can
> rerun cmake and submit the cmake files on a system with boost 1.46 if
> needed.

Maybe delete CMakeCache.txt and rerun cmake to make sure it refreshes
everything.

> I have uninstalled the 1.46 libs manually (except those that are
> required by some applications) and the installed libboost-xyz1.48-dev
> libraries. recompiled uhd and gnuradio and it seems the run-time issue
> I was having before (app freezes) is now gone.
>
> Alex

Ok, glad you got it working.

Tom



reply via email to

[Prev in Thread] Current Thread [Next in Thread]