[Nottingham] Fwd: Microsoft & Github

VM vadim+NLUG at mankevich.co.uk
Fri Sep 16 10:23:38 UTC 2016


Forwarding this to the group.
>> On 16 Sep 2016, at 09:02, Roger Light <roger at atchoo.org> wrote:
>> Now the other side of this is that there may have been 10,000+ people
>> providing patches but that could be down to the original code not
>> being of high quality rather than just being of interest to a lot of
>> people. Either way, the code has most likely got a lower density of
>> bugs than it did before Microsoft put it on github.
> 
> That is true. It was not of high quality to start with and in my opinion beyond repair. Tens of thousands contributors don't really fully understand how the damn thing was supposed to work. Thus, whether software becomes better or worse depends on a few core developers who can absorb contributions and integrate them bearing in mind the big picture. They failed on that in the first place. Will they perform better with larger flow of disconnected contributions?
> I read a couple of pages of comments on github - the discussion of how powershell was deliberately made incompatible with Linux. Unbreaking it is a destructive change and Microsoft would not accept patches to correct it straight away.
> I do believe in open source and community development but right from the start of a project. If the project lives long (think Linux) there will be a competent quality assurance team and development workflow that enables huge amount of contributions to very complex software that still becomes better (at least not worse) with every major release.
> Just borrowing a workflow tool (git) and philosophy for which that tool was designed doesn't miraculously make a previously closed-source project a success.
> And in order not to be bashing only Microsoft I think that "quantity is not quality" also applies to major open source projects (e.g., xen, systemd) but from a different perspective. They are just too critical to be large and complex.
> 
> VM



More information about the Nottingham mailing list