[Rivet] PATCH: Don't re-download boost tarball if re-running the bootstrap script

[Andy Buckley]

Thanks for the fix... it's gone public now.

For the warning silencing, I'm not a compiler warning guy either, but if
the warning mentions "[ -Wunused-local-typedefs ]" then it's a good bet
that adding -Wno-unused-local-typedefs to the compiler command will shut
it up ;-)

Andy


On 10/06/14 22:07, David Bjergaard wrote:
> Hi Andy,
> 
> Your method is much cleaner than mine.  I was working from google
> searches and bug reports that I cobbled together.  I'm not a g++
> compiler flag expert.  (And patching is a pain)
> 
> Unfortunately I think you wanted:
>     test -f boost_$BOOST_VERSION.tar.gz || wget http://downloads.sourceforge.net/boost/boost_$BOOST_VERSION.tar.gz
> instead of
>     test -d boost_$BOOST_VERSION.tar.gz || wget http://downloads.sourceforge.net/boost/boost_$BOOST_VERSION.tar.gz
> 
> This will still re-download the tarball even if it already exists, since
> the tarball is not a directory.  Even more worrisome is the case where a
> user makes a directory named boost_$BOOST_VERSION.tar.gz, skipping the
> download and causing the script to crash and burn later on... I know I'm
> being a little facetious, but it is possible.
> 
>     Dave
> 
> Andy Buckley <andy.buckley at cern.ch> writes:
> 
>> Hi David,
>>
>> The patch for Boost looked a bit heavy to me (maybe I misread the patch
>> file but it looked like a code adaptation) but I think it is made
>> acceptable by setting this in the bootstrap script:
>>
>> export CXXFLAGS="$CXXFLAGS -Wno-unused-local-typedefs"
>>
>> This disables the noisy warning and *should* propagate to
>> rivet-buildplugin since the latter contains this code to work out what
>> flags to build plugins with, based on how Rivet itself was built:
>>
>> ## Get Rivet system C++ compiler flags
>> mycxxflags=""
>> if [[ -n "@AM_CXXFLAGS@" ]]; then
>>     mycxxflags="@AM_CXXFLAGS@"
>> fi
>> if [[ -n "@CXXFLAGS@" ]]; then
>>     mycxxflags="$mycxxflags @CXXFLAGS@"
>> fi
>>
>> Of course I only did this when I updated my own system and started
>> noticing the problem ;-)
>>
>> Thanks again for the report,
>> Andy
>>
>>
>> On 16/05/14 20:52, David Bjergaard wrote:
>>> Hi Andy,
>>>
>>> If you're interested... there's a "bug" in BOOST 1.55.0 that causes a
>>> bunch of warnings to be printed in gcc 4.8. I know this is upstream, and
>>> it will be fixed in the next BOOST release, but in the meantime the
>>> rivet-buildplugin script is very loud.  
>>>
>>> I've got a patch that suppresses all of the warnings:
>>> https://github.com/dbjergaard/rivet-charmonium/blob/master/boost_1_55.patch
>>>
>>> I've attached a patch file that may work for the bootstrap, though I
>>> think I may have structured the patch for:
>>> INSTALL_DIR=~/rivet/local
>>> BUILD_DIR=~/rivet
>>>
>>> If you want, I can rework it to patch before the headers get installed. 
>>>
>>>     Dave
>>>     
>>>     
>>>
>>>
>>>
>>>
>>>
>>> Andy Buckley <andy.buckley at cern.ch> writes:
>>>
>>>> Thanks David! Actually the bootstraps could do with some general
>>>> protections when re-running... will add.
>>>>
>>>> Andy
>>>>
>>>>
>>>> On 15/05/14 17:20, David Bjergaard wrote:
>>>>> Hi,
>>>>>
>>>>> Here's a patch to avoid unnecessarily downloading multiple copies of the
>>>>> boost tarball if the rivet-bootstrap script fails for some other reason:
>>>>>
>>>>> 75c75
>>>>> <     wget http://downloads.sourceforge.net/boost/boost_$BOOST_VERSION.tar.gz
>>>>> ---
>>>>>>     [ ! -f boost_$BOOST_VERSION.tar.gz ] && wget http://downloads.sourceforge.net/boost/boost_$BOOST_VERSION.tar.gz
>>>>>
>>>>> The bootstrap script has much improved since 1.8! Thanks for all the
>>>>> hard work.
>>>>>
>>>>> Cheers,
>>>>>
>>>>>     Dave
>>>>> _______________________________________________
>>>>> Rivet mailing list
>>>>> Rivet at projects.hepforge.org
>>>>> https://www.hepforge.org/lists/listinfo/rivet
>>>>>


-- 
Dr Andy Buckley, Royal Society University Research Fellow
Particle Physics Expt Group, University of Glasgow / PH Dept, CERN