Bug#794915: comments regarding your review of compute

[Gianfranco Costamagna <costamagnagianfranco@yahoo.it>]

Hi Ghislain,

>I did not expect a review so soon, many thanks!


sorry to ruin your Saturday :)


>Should be arch:all indeed.


ack
>That one is a bit difficult. It is temporarily disabled because at the 
>time v0.4 was released the test suite was quite buggy and most tests 
>failed. According to the upstream git log, a few fixes have landed in 
>the development branch, which now makes most, but not all, tests pass
>
>I am working with Kyle to fix the remaining failures of the test suite 
>for a future release.
>
>Meanwhile, I tested the package on the test suite of the OpenCL backend 
>of ArrayFire (for which Boost.Compute is a dependency) and it all 
>passes. So, I have reasonable confidence that the Boost.Compute code 
>actually works and that it is worth shipping in a package today.


ack for this, just please remember to add them in a future release :)
(if possible)

or you might think about shipping the tests that actually pass with a patch
and remove that patch in a future release (this way you will remember easily
about my words :D)

>You are absolutely right. This is more coming from my "explicit better 
>than implicit" school of thoughts.


fine for me


>Yes, Expat should be compatible with BSL-1.0, since the wording is the 
>same minus the added clause in the middle.


yes, indeed
>I usually don't take any risks and just let licensecheck do its magic here.

> Plus, it took me a while ^^.


I do the same, so fine for me

>If you're happy with my comments, I intend to:
>   * fix 1) now,
>   * fix 2) later, when the test suite is properly fixed upstream,


I will trust your memory there :)

>   * let 3) and 4) as is.>
>And push a new version to mentors.
>
>Please let me know how that sounds.


sounds reasonable.


Ping me when the new release is up and I'll review it again,

thanks,

Gianfranco