Re: [PATCH 1/1] ncurses: Update to 5.9

[Tom Rini <tom_rini@mentor.com>]

On 04/06/2011 11:32 AM, Khem Raj wrote:
> On Wed, Apr 6, 2011 at 11:29 AM, Tom Rini <tom_rini@mentor.com> wrote:
>> On 04/06/2011 11:27 AM, Khem Raj wrote:
>>> On Wed, Apr 6, 2011 at 10:35 AM, Tom Rini <tom_rini@mentor.com> wrote:
>>>> On 04/06/2011 10:26 AM, Khem Raj wrote:
>>>>> On Wed, Apr 6, 2011 at 10:10 AM, Tom Rini <tom_rini@mentor.com> wrote:
>>>>>> On 04/06/2011 10:05 AM, Khem Raj wrote:
>>>>>>> On Wed, Apr 6, 2011 at 7:30 AM, Tom Rini <tom_rini@mentor.com> wrote:
>>>>>>>> On 04/05/2011 11:18 PM, Khem Raj wrote:
>>>>>>>>> On Tue, Apr 5, 2011 at 5:38 PM, Tom Rini <tom_rini@mentor.com> wrote:
>>>>>>>>>> The previous 5.7 release was relatively close to 5.8 due to it bringing
>>>>>>>>>> in a patch to sync with upstream work-in-progress.  We skip over the
>>>>>>>>>> 5.8 release and move to 5.9.
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> there already are patches for 5.9 available too
>>>>>>>>> ftp://invisible-island.net/ncurses/5.9/ncurses-5.9.patch.gz
>>>>>>>>
>>>>>>>> Wrong link?  That reverse applies to ncurses 5.9 release. But
>>>>>>>> regardless, is ncurses something we need to be tracking top of tree for?
>>>>>>>>  It seems like we needed to for 5.7 since there had been a lot going on
>>>>>>>> without a release but that seems to have changed now.
>>>>>>>>
>>>>>>>
>>>>>>> those patches usually contain critical bug fixes including security updates
>>>>>>> so it will be of interest to keep track of it
>>>>>>
>>>>>> Well, it doesn't currently.  And while I agree we need to do a good job,
>>>>>> everywhere, of keeping track of security updates, I don't think we
>>>>>> should move back to depending on a site that frequently removes patches.
>>>>>>
>>>>>
>>>>> yes. cache the patches like yocto did for 5.7 recipes
>>>>
>>>> That still leaves the problem of there not being a valid patch there at
>>>> the moment.  And I still don't see why ncurses needs to be in the bucket
>>>> of recipes we track the scm for rather than relying on the latest stable
>>>> release.
>>>
>>> 5.9 was released few days back so that patch might be lean for now
>>> but I assume overtime it will get fatter
>>
>> It's invalid at the moment, yes.  But you haven't explained why ncurses
>> needs to be in the bleeding edge bucket.  Usually this is for stuff that
>> hasn't really reached a stability point.
>>
> It does not have to be but those patches are cumulative fixed that are done
> on top of a release. I am sure we will also run into the problems those will
> fix thats why its better to keep and eye on them

That can be said for just about every recipe we have.  It sounds like
you're suggesting we need _svn recipe or similar recipe for ncurses as
well.  I still don't see why ncurses is special in this regard and ask
that when you see a worthwhile patch for ncurses 5.9 that you do another
pull request.  I'm just trying to keep oe-core in sync with
openembedded.master.

-- 
Tom Rini
Mentor Graphics Corporation