Re: [Open64-devel] cleanup patch to remove PROMPF code [LNO+]

[David C. <dco...@gm...>]

Christopher, thanks for the input, I agree with your reasons for
removing the code.

Have any of the Open64 gatekeepers looked at this change yet?

2012/2/27 "C. Bergström" <cbe...@pa...>:
> On 02/27/12 09:23 AM, David Coakley wrote:
>>
>> To answer a few questions:
>>
>> Q: What is PROMPF?
>> A: It is a static analysis feature which generates an additional file
>> (.anl) during compilation.  There is a lot of code in LNO to maintain
>> related information in WHIRL, but I don't think it is used to make any
>> changes during compilation.  I believe the expectation was that
>> another tool would be used to interpret the analysis file... maybe
>> someone from the SGI days can clarify.
>
> PathScale in the coming months may make an external tool which can read the
> .anl files.  I believe UH also had a project that could read it, but I'm not
> sure the status.  We're happy to work with the community on improving this
> functionality and documenting any changes we make to the .anl format
>
>>
>> Q: Why does the patch remove the Whirl Browser?
>> A: It does not.  It only removes the WB code related to PROMPF.
>>
>> Q: Why not maintain or improve this functionality?
>> A: The patch is based on the assumption that no one is using this
>> feature because BUILD_SKIP_PROMPF has always been defined for all
>> targets in Open64.  I tried to build a compiler with PROMPF support on
>> x86 and the feature didn't work, because no prompf_anl.so is created
>> and copied to the installation directory, further convincing me that
>> no one is using the feature.
>
> It should be fairly trivial to get this building.
> -----------------------
> TBH I'd support removal of this code since
> 1) I think this functionality can be improved and the current state may not
> be the best place to start
> 2) Unless the Open64 community is willing to develop their own or
> collaborate with others you're not guaranteed to be able to use the external
> tools