On Thu, Dec 8, 2011 at 9:58 PM, Lex Berezhny <[email protected]> wrote:


> Something tells me that the repository for cpython is much larger than
> pyjamas and will likely always be so.

 that's not a reason to impose a performance hit, for all time, across
hundreds or thousands of developers, the sum of which is vast amounts
of time and energy wasted.

 to whit: mercurial is not, cannot be, and never will be as good as
git, performance-wise nor probably feature-wise, although i will never
know about feature-wise because i won't ever use it (other than to
check out code which cannot be accessed any other way) because the
primary priority - performance - can never be reached.

> I say if it's good enough for python it should be good enough for pyjamas ;-)


> Don't forget Mozilla! There is a lot more code being managed by
> Mozilla than by Linus.

 i didn't say linus, did i?  have you seen the size of the code in
pythonwebkit?  also, the trinitydesktop team are just transferring
everything over to git, right now.  importing the history from svn is
a bitch...

> Saying that mercurial is somehow not capable enough because it's
> written in Python when several very large visible projects and
> thousands of smaller projects use it seems unfair.

 it's "unfair" only if you believe that just because the lead
developer of pyjamas happens to have picked git that somehow that
developer is somehow an authority on the subject of suitability of
mercurial for those very large visible projects and thousands of
smaller projects.

 ... they're not, are they?  i don't have commit rights to any of
those thousands of projects, nor am i an authority in those
communities?

 in other words, those projects are entirely free to choose that which
they prefer to experience.

 i prefer to experience greatly-reduced update times, commit times,
ease of branching, massively-reduced switching time between branches
and highly-optimised greatly reduced push and pull network transfer
times.

 i'm not sure if you're aware of this, but git checkouts are actually
quicker than doing "cp" of the exact same set of files (or even an
untar of a compressed archive).  the system-level experience of the
developers behind git allows them to make some serious, serious
optimisations.

 i'm sure that someone will go and prove that that's wrong, but hey... :)

 l.

Reply via email to