[eigen] Re: Scheduling beta3 on sunday february 6 ?

[Benoit Jacob <jacob.benoit.1@xxxxxxxxx>]

It's also time to start testing the heck out of it.

http://eigen.tuxfamily.org/index.php?title=Tests

It is specially useful to test on exotic (read: non-GCC,
non-recent-GCC, non-x86) configs, and to try with these CMake options:

cmake -DEIGEN_DEFAULT_TO_ROW_MAJOR=ON

   tests with RowMajor as the default storage order (this one is most
important! some users rely heavily on this, yet it doesn't get enough
testing).

cmake  -DEIGEN_TEST_OPENMP=ON

   tests OpenMP support

cmake -DEIGEN_TEST_EIGEN2=ON

   tests the eigen2 tests against EIGEN2_SUPPORT

Benoit

2011/2/3 Benoit Jacob <jacob.benoit.1@xxxxxxxxx>:
> Hi,
>
> Here's a new update. Beta3 is just a couple of MSVC fixes away!
>
> http://eigen.tuxfamily.org/bz/showdependencytree.cgi?id=48&hide_resolved=1
>
>       156: Several CDash Tests fail under MSVC 2008 + 2010
>
> This was posted 2 weeks ago. Does the problem still exist?
>
>       165: Inverse unit test crashes (release mode/msvc)
>
> This one is confirmed.
>
> On my side I'll be working on updating documentation. If you want to
> help, there's plenty of work to do on that front. For example, you may
> have noticed that the tutorial links to several special-topic pages,
> e.g. on resizing, that haven't been written yet. Feel free!
>
> Benoit
>
>
> 2011/1/27 Benoit Jacob <jacob.benoit.1@xxxxxxxxx>:
>> Hi,
>>
>> Thanks to everybody's hard work, beta3 already looks much closer today!
>>
>> http://eigen.tuxfamily.org/bz/showdependencytree.cgi?id=48&hide_resolved=1
>>
>>       51: no malloc in fixed-size products
>>
>> -> for Gael; see my previous email
>>
>>       52: Strides API
>>
>> -> for me; see my previous email
>>
>>       53: Sparse - API stability for basic features
>>
>> -> for Gael; see my previous email
>>
>>       57: Uniform external library support
>>
>> -> for Gael; see my previous email
>>
>>       73: 3.0-beta2: Intel incomplete type error breaks several unit tests
>>
>> -> still needs attention
>>
>>       100: Finalize the array-as-scalar feature
>>
>> -> see my previous email
>>
>> Benoit
>>
>>
>> 2011/1/26 Benoit Jacob <jacob.benoit.1@xxxxxxxxx>:
>>> Hi,
>>>
>>> Despite the flurry of improvements currently happening in hg, we still
>>> need to focus and actually release this beta3 !
>>>
>>> Let's schedule it. Do you agree for Sunday February 6 ? That leaves us 10 days.
>>>
>>> That means that we have at most one week to finish it. Does that sound
>>> OK? The blockers are:
>>>
>>> http://eigen.tuxfamily.org/bz/showdependencytree.cgi?id=48&hide_resolved=1
>>>
>>>       51: no malloc in fixed-size products
>>>
>>> Gael, if you don't have time, this can be dropped from beta3 and block
>>> only final.
>>>
>>>       52: Strides API
>>>
>>> Will do (if I make API changes there, i'll write to this list very
>>> soon with my proposal)
>>>
>>>       53: Sparse - API stability for basic features
>>>
>>> Gael, please decide very soon the fate of Sparse in 3.0! As we
>>> discussed, even if it's not stable, we dont necessarily have to move
>>> it to unsupported: it is rather less intrusive to force users to
>>> #define I_KNOW_THAT_SPARSE_IS_NOT_YET_STABLE.
>>>
>>>       57: Uniform external library support
>>>
>>> Gael -> if Sparse isn't stable in 3.0, then this is no longer a blocker..
>>>
>>>       59: Clean up FullPivHouseholderQR
>>>
>>> Will do very soon.
>>>
>>>       73: 3.0-beta2: Intel incomplete type error breaks several unit tests
>>>
>>> This one needs attention. Can anyone install ICC and have a look?
>>>
>>>       100: Finalize the array-as-scalar feature [See dependency tree
>>> for bug 100]
>>>
>>> Shouldn't we just make this a 3.1 feature at this point? How does it
>>> affect existing API?
>>>
>>>       131: Unaligned Quaternions/Transformations
>>>
>>> Need to have a look at this. I think it was added recently.
>>>
>>>       148: array-matrix transpose product does not compile
>>>
>>> Same.
>>>
>>> Benoit
>>>
>>
>