Update on dashboards, testing and CMake

4 views
Skip to first unread message

Marcus D. Hanwell

unread,
Oct 10, 2010, 2:30:34 PM10/10/10
to ryppl-dev
Hi,

After my international travels I finally found a little time to update you on some progress. I have updated the dashboard builds, adding a new Windows XP vs9 submission, probably a Mac submission and some coverage reports using gcov.


I will be taking a look at the header generation stuff, and have some ideas on how to fix this and get rid of a bug on Windows where the source and build are on different drive letters. The coverage is showing as 88.1% with 257 tests not running. I still have 207 tests failing, and on Windows there seems to be an additional issue with versioned library names that I will need to look into. This is where most of the compile errors on Windows are coming from - it looks like the version is not being set when the targets are created. I will do some digging.

It would be great to sync up at some point to ensure I am not about to duplicate anything Eric is working on. 

Marcus D. Hanwell, Ph.D.
R&D Engineer, Kitware Inc.
(518) 881-4937

Eric Niebler

unread,
Oct 10, 2010, 3:13:40 PM10/10/10
to rypp...@googlegroups.com

Hi Marcus,

Welcome back! I'm very glad to see you getting involved in ryppl again.
I have been keeping the modularized boost in sync with trunk, but I
haven't made much additional progress beyond that. The project has
reached a lull, blocked on CMake support on one side and dependency
resolution on the other. I'd be happy to actively work with you to
improve CMake support for modularlized boost. Personally, I'd like to
see us push git+cmake+modularization forward and try to transition boost
to that asap. There are no technical obstacles there, so it seems doable.

Any other suggestions for plans to proceed?

--
Eric Niebler
BoostPro Computing
http://www.boostpro.com

signature.asc

David Abrahams

unread,
Oct 10, 2010, 7:28:22 PM10/10/10
to rypp...@googlegroups.com
At Sun, 10 Oct 2010 12:13:40 -0700,

Eric Niebler wrote:
>
> On 10/10/2010 11:30 AM, Marcus D. Hanwell wrote:
> > Hi,
> >
> > After my international travels I finally found a little time to update
> > you on some progress. I have updated the dashboard builds, adding a new
> > Windows XP vs9 submission, probably a Mac submission and some coverage
> > reports using gcov.
> >
> > http://my.cdash.org/index.php?project=Ryppl
> >
> > I will be taking a look at the header generation stuff, and have some
> > ideas on how to fix this and get rid of a bug on Windows where the
> > source and build are on different drive letters. The coverage is showing
> > as 88.1% with 257 tests not running. I still have 207 tests failing, and
> > on Windows there seems to be an additional issue with versioned library
> > names that I will need to look into. This is where most of the compile
> > errors on Windows are coming from - it looks like the version is not
> > being set when the targets are created. I will do some digging.
> >
> > It would be great to sync up at some point to ensure I am not about to
> > duplicate anything Eric is working on.
>
> Hi Marcus,
>
> Welcome back! I'm very glad to see you getting involved in ryppl
> again.

Hear hear!

> I have been keeping the modularized boost in sync with trunk, but I
> haven't made much additional progress beyond that. The project has
> reached a lull, blocked on CMake support on one side and dependency
> resolution on the other. I'd be happy to actively work with you to
> improve CMake support for modularlized boost. Personally, I'd like
> to see us push git+cmake+modularization forward and try to
> transition boost to that asap. There are no technical obstacles
> there, so it seems doable.
>
> Any other suggestions for plans to proceed?

I'm going to give my status update in a separate message. I was just
writing it when I saw Marcus's message come in!

--
Dave Abrahams
BoostPro Computing
http://www.boostpro.com

Marcus D. Hanwell

unread,
Oct 10, 2010, 7:54:47 PM10/10/10
to rypp...@googlegroups.com
This seems like a reasonable path forward, and one I would be in favor of. I will chase up the Mac dashboard tomorrow, and see what I can do to close up some of the bug reports you opened up for me. I will also be around on IRC this week if you need IRC. It would be great to get things in shape so that we can start using it in some of our projects too.

Thanks,

Marcus

Eric Niebler

unread,
Oct 19, 2010, 6:26:33 PM10/19/10
to rypp...@googlegroups.com
On 10/10/2010 4:54 PM, Marcus D. Hanwell wrote:
> This seems like a reasonable path forward, and one I would be in favor
> of. I will chase up the Mac dashboard tomorrow, and see what I can do to
> close up some of the bug reports you opened up for me. I will also be
> around on IRC this week if you need IRC. It would be great to get things
> in shape so that we can start using it in some of our projects too.

Hi Marcus, a few questions. Have you grokked enough of the modularized
boost cmake system to share what you've learned? Is there something I
can do to help out with these cmake-related bugs?

signature.asc

Marcus D. Hanwell

unread,
Oct 24, 2010, 1:43:25 PM10/24/10
to rypp...@googlegroups.com
Hi Eric,

One of the larger outstanding issues, which I think I talked to you
about previously, is the missing name resolution in the build system.
So if I want the static form of the library, rather than the default
shared form, it seems to be missing.

I also need to hunt down what happened to the version components in
some of the libraries on Windows, where the version is encoded in the
library name. This was working, and it looks like the CMake variables
are empty. I think I grokked the majority, and am working on setting
aside some time to write about what I have learned, and fix some of
these bugs that have turned up.

Having nightly builds on all three major platforms helps. I will make
myself more available on IRC. I am also eager to start using this form
of Boost in some of the projects I work on, as it would be far more
convenient for us and would provide more testing feedback.

Marcus

Reply all
Reply to author
Forward
0 new messages