more performance work?

[Vincent Ladeuil]

>>>>> "bialix" == Alexander Belchenko <bialix at ukr.net> writes:

    bialix> James Westby пишет:
    >> On (08/11/07 22:41), Andrew Cowie wrote:
    >>> On Thu, 2007-11-08 at 00:01 +1100, Robert Collins wrote:
    >>>> Are there obvious things I'm missing, in terms of thinking about where
    >>>> we should be putting our energies?
    >>> I wouldn't dream of suggesting that performance shouldn't have been the
    >>> top priority for the last few months, and your progress there has been
    >>> fantastic. Chatter in #bzr, along with various minor but long standing
    >>> bugs, however, somewhat indicate that usability and Just Works factor
    >>> has fallen behind the curve a bit.
    >>> 
    >>> Since "better user experience" is ultimately Bazaar's claim to
    >>> distinction, I would encourage those contributing to the code to
    >>> consider potentially giving some consideration toward taking your
    >>> already awesome tool and further refining its friendliness, simplicity,
    >>> conciseness, and ease of use.
    >>> 
    >> 
    >> I agree.
    >> 
    >> There are also some bugs that while not critical for everybody, or with
    >> easy workarounds are severely hampering some users.
    >> 
    >> I would support delaying 1.0 in order to spend time addressing bugs and
    >> polishing up. With the announcement of a 1.0 release many people may try
    >> out bzr, and if there are many known bugs they can trip over they will
    >> have a bad impression, and could lead to plenty of bad press.

    bialix> +1.

    bialix> Releasing 1.0 now is a bit too fast for me.

I'm both +1 and -1 on releasing 1.0

Minus side:

- too many bugs on launchpad, and the number is rising with each
  release, which means that we don't spend enough time fixing
  them. More exposure will make the problem worse,

- there are still room for improvements in many areas (including
  doc, better import from foreign VCS, etc.)

Plus side:

- let's face it, this tool Works,

- 1.0 doesn't mean perfect,

- bzr still has the reputation of being slow while in the some
  time a lot of people keep coming to try it and keep using it,
  let's get more people trying it and we will surprise them with
  further performance improvements :)

- we *need* more exposure to stress test and get a larger
  feedback on what was accomplished until now.

So all in all :-)

   Vincent