Anyone still using smart ?
Anders F Björklund
afb at algonet.se
Thu Aug 11 08:38:57 PDT 2016
11 aug 2016 kl. 17.24 skrev Mark Hatle:
> On 8/11/16 10:11 AM, Jeffrey Johnson wrote:
>>
>>>> I have mostly received requests to redo the whole thing in Py3 or Qt4, to no
>>>> effect.
>>>> (i.e. it was ported from python to python3, or from gtk to qt4. but still the
>>>> same)
>>>>
>>>> Should we keep the project alive ? Does anyone want to take over the
>>>> maintenance ?
>>>
>>> We are still using smart in the Yocto Project. We are starting to look into
>>> DNF, but it won't happen before spring of next year... If the requirements or
>>> integration w/ RPM5 are too difficult, we will be sticking with smart until we
>>> find a replacement.
>>>
>>
>> Using DNF is feasible, just the issues are similar to smart vis-a-vis Py3 or Qt4:
>>
>> Ports aren’t all that exciting, particularly when the result resembles Frankenstein.
>>
>>> Frankly smart works, and works well for our embedded use cases... so we've got
>>> more important pieces to work through first. (Transition to Python3 is what is
>>> driving the look into DNF and others.)
>>>
>>
>> What is the issue with Python3 porting? Does the tool that claims to automate
>> python2 -> python3 not work that well?
>
> We need someone to go through and verify it works after running through such a
> tool... and then we've still got a support and 'marketing' problem. Support
> being we've got limited resources to keep things running going forward. (There
> are about 4 or 5 of us, each part time, keeping smartpm [and the YP build
> system] working with the Yocto Project use cases.) The marketing problem is
> 'but everyone else uses XYZ, why don't you?!'.
My main problem with the python port, was that there no (feasible) way that it
could support *both* of python and python3 at the same time. Requiring that all
maintenances releases would install "python2" (2.7) was not really realistic.
But the code is still available, of course:
https://github.com/smartpm/smart/tree/py3k
It required porting both the regular python code (2to3) and the C code (2to3c)
I think that it mostly worked, but then again that was back in 2012 anyway...
Never did any attempt to do a gtk3 port, but I suppose that could be "needed".
Here was the most recent request for py3:
https://github.com/smartpm/smart/issues/4
I'm fine with keeping the current project structure alive, was just curious. :-)
But probably won't have any development time to spend on it myself, this year.
Was more fishing for a roadmap, while I was looking over my github repositories.
The smart repository was moved to a group, so it could be co-maintained if needed:
"Smart Package Manager Team"
https://github.com/smartpm
--anders
More information about the Smart
mailing list