On Sun, Oct 20, 2002 at 12:32:43AM -0700, Ian Clarke wrote:
> On Sat, Oct 19, 2002 at 11:15:38PM -0400, Michael Wiktowy wrote:
> > >So whenever a bug is discovered during that time, the clock is reset to
> > >1 month from that time? I doubt that very much.
> > 
> > I certainly didn't interpret a week-countdown reset each time a bug is found
> > in Oskar's bug release pseudo-code. That would be crazy! We would never
> > ever see a point release. What I am saying is to model the release after 
> > Mozilla
> > point releases. Freeze feature additions for a short period of time and 
> > focus on
> > killing show stopper bugs rather then feature additions and when a build 
> > looks
> > stable rebadge it as the point release. I would say that 0.5pre5 meets 
> > that goal.
> 
> Exactly, that is what we have been doing since 0.5pre1 several weeks 
> ago.
> 
> > The average user not being able to install on Windows would be a
> > blocker, no? If a good Windows installer could be built around
> > 0.5pre5 then I would vote for that being 0.5final having used/tested
> > every pre release so far. The snapshots after that point have become
> > unstuck again.
> 
> With all due respect to those working on the Windows installer, I am
> simply saying that the Windows installer is so simple that it blocking 
> the 0.5 release would be rediculous, and would only result in a total 
> lack of motivation on the part of those people.  Fortunately, those 
> working on the Windows installer show no-such lack of motivation.
> 
Releasing 0.5 without it would totally destroy our credibility in the
eyes of the press and some of the public, that we are trying to
convince.
> > Netscape was beta (at least it was taken from pre 1.0 mozilla code) and
> > pawned off as stable. That is what did more harm then good. Whether
> > you like it or not, the perception is there that this next release of 
> > Freenet
> > will be functional and stable. You would be fostering that perception
> > by calling the focus of journalists on this release ... the first point 
> > release
> > after more the a year. Don't you think that they are going to be expecting
> > great things? The least that we can do is file off the pointy edges so
> > they don't take an eye out when they come to look. How many will be
> > running Linux on their computers that they want to try it out on?
> 
> We have been "filing off the pointy edges" since 0.5pre1.  Someone has 
> to draw a line in the sand, it isn't a popular position to be in, but as 
> the project coordinator, I need to do it.
We need to file off the pointy edges just a little more. A trivial patch
to be applied shortly will reduce worst case startup times by about 2/3,
but with large stores it might still be too long. Fixing the rest means
a not-quite-trivial modification to NativeFSDirectory (making the hash,
and the LRU, per-dirID). The reseed-every-run bug also needs to be dealt
with because it compromizes network performance and if we deploy 10,000
nodes with it it will be a huge loss. Thus there will be a new prerelease
with significant new code. There is also a rare LRU corruption bug that
I will find with some systematic testing probably, but it's nothing to
worry about. Thus, we will have a prerelease on monday or tuesday, and
final on wednesday... this isn't very long for testing such
modifications.
> 
> > No one is recommending that people should use 0.3. 
> 
> Actually, we *are* implictly recommending that people should use 0.3
> since 0.3 is the current stable release.
Hmmm. Are we? The prereleases are better than 0.3, and this point should
be made prominently on the website.
> 
> > What is fair to Freenet users is:
> > - having some sort of release process and schedule
> 
> Yes indeed.  A schedule works both ways.  It forces developers to adhere 
> to a release schedule that makes them sweat a-little bit (as I have 
> tried to do), while not giving them a totally impossible task.
> 
> > - having an effective means of bug reporting that is not ignored by devs
> 
> devl at freenetproject.org
> 
> > - having devs eat a bit of their own dogfood
> 
> I test freenet extensively on a daily basis.
> 
> > - having well documented installation routines or easy installers
> 
> Both the linux and windows installers are pretty easy to get to grips 
> with.  If you don't think so, I would appreciate your constructive 
> criticism.
> 
> > - implementation of features that users are begging for
> 
> Of course, but these do not have to be implemented before 0.5, there 
> will be a 0.5.1, and a 0.5.2 etc in which such features can be provided.
Ian is right here. Apart from the proviso that not all features that
users want are practical or sensible.
> 
> > So far the Freenet Project does not score well on the above points but
> > it is getting better and better. Matthew is one of the bright points in
> > this whole process that leads me to hope for the future of Freenet.
> 
> I agree, and please don't forget that the reason Matthew has been able 
> to do what he has done is because he is being paid from the very 
> donations, that result from the publicity that so many have been 
> busy sneering at in recent days.
The big build up is precisely the reason that 0.5 has to be of
acceptable quality.
> 
> Ian.
> 
> -- 
> Ian Clarke                ian@[freenetproject.org|locut.us|cematics.com]
> Latest Project                                 http://cematics.com/kanzi
> Personal Homepage                                     http://locut.us/



-- 
Matthew Toseland
toad at amphibian.dyndns.org
amphibian at users.sourceforge.net
Freenet/Coldstore open source hacker.
Employed full time by Freenet Project Inc. from 11/9/02 to 11/11/02.
http://freenetproject.org/
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
URL: 
<https://emu.freenetproject.org/pipermail/devl/attachments/20021020/56376b88/attachment.pgp>

Reply via email to