[Halld-offline] Fwd: updates to branch sim-recon-rj-pm checked in

[Richard Jones]

Mark,

If you could back out the change, then go through and remove just the
Makefiles that coexist with SConscript files in the same directory, that
should be easy to do. Then as we move forward with the conversion to scons,
we can progressively remove the Makefiles and replace them with SConscript
files. This prevents us from having to maintain two build systems
simultaneously, while allowing the transition to scons to be incremental.

Meanwhile, I have checked changes into the BMS area on my working branch
(sim-recon-rj-pm) that makes the BMS system continue to work under the new
hierarchy for include/bin/lib dirs for those areas of the build tree that
are not converted to scons. You might want to move them over to the trunk
(changes to BMS), for people who work with tools that are not currently
instrumented for building with scons.

Can someone remind me why the decision was made to switch the build system
for sim-recon over to scons before the conversion from make was complete?

-Richard J.


On Mon, Jul 28, 2014 at 2:41 PM, Mark M. Ito <marki at jlab.org> wrote:

> Richard et al.,
>
> On 07/26/2014 01:57 PM, Richard Jones wrote:
>
>> Hello all,
>>
>> ... Also a note to Mark, I hope we did not remove the Makefiles from the
>> branches of the tree where scons has not yet been implemented.
>>
> I'm afraid we did (on the trunk).
>
>  There are useful tools out there that can only be built using make, as of
>> today. For example, consider src/libraries/AMPTOOLS_DATAIO. When I need one
>> of these, I simply apply whatever fixes to the BMS system is needed to
>> restore the former build functionality. If the Makefiles are globally
>> stripped then that leaves whole branches of sim-recon without any build
>> system at all, which is probably a step backward. It should probably be
>> either make or scons, but not both, and not neither.
>>
> So my impression is that we were not ready for a global makefile removal.
> We need to be more targeted. Identify the special directories, restore make
> in those directories only or convert them to scons.
>
> Sounds like right now the trunk has broken directories, i. e., there is no
> way to build them right now. The quickest thing is to reverse my changes
> from Friday completely, i. e., put the makefiles and BMS back the way they
> were before, and have a better plan for going forward next time. That is my
> proposal; comments?
>
>   -- Mark
>
>
> --
> Mark M. Ito, Jefferson Lab, marki at jlab.org, (757)269-5295
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.jlab.org/pipermail/halld-offline/attachments/20140728/182f3cb4/attachment-0002.html>