WebOb 1.1beta1 released
Sergey Schetinin
Alex Grönholm
> http://pypi.python.org/pypi/WebOb/1.1beta1
> http://docs.webob.org/en/latest/news.html#beta1
>
Get that thing off of PyPI please, right away. Everyone who
automatically installs webob as a dependency gets this beta.
http://tarekziade.wordpress.com/2011/02/15/do-not-upload-dev-releases-at-pypi/
Alex Grönholm
> http://pypi.python.org/pypi/WebOb/1.1beta1
> http://docs.webob.org/en/latest/news.html#beta1
>
the changelog. Is the project still on track for that?
Chris McDonough
I agree that it would be far better for PyPI to not advertise by default
to pip/easy_install releases that "look like" alphas and betas. But I
think it's caveat emptor when someone uses PyPI at all; you can't really
depend on it for any sane result at any given time (you need to create a
private index for that).
- C
Chris McDonough
http://lists.repoze.org/pipermail/gsoc-pylons-2011/ is the best place to
keep an eye on for this, likely.
- C
Sergey Schetinin
Yeah, even though http://pypi.python.org/pypi/WebOb/json seems to be
smart enough to pick the non-beta version, setuptools / distribute are
not. I had both 1.0.8 and 1.1beta1 shown, but I recon some people
might be caught by surprise still. (I don't think there's anything
questionable in the beta though, so it's not like their code would
break, unless they use python2.4 or req.postvars etc). Anyway, I've
hidden the beta release.
Chris McDonough
And please, don't get this thing off PyPI. The effects of deleting a
release lead to far worse results than just leaving it there and
releasing a 1.0b2 quickly if there turn out to be bugs or bw incompats.
Sergey Schetinin
I don't intend to delete it, just set it to hidden, so that
http://pypi.python.org/pypi/WebOb shows 1.0.8 instead of a choice
between 1.0.8 and 1.1beta1. I'm not sure if hiding is the right thing
to do.
So I suppose Alex is +1 for hiding
I'm -1, cause it's barely a release at all if it's not on PyPi.
Chris?
Chris McDonough
Insanely, hiding it in the PyPI UI doesn't mean pip and easy_install
don't find it because it's still listed on
http://pypi.python.org/simple/WebOb/ (which is the default index page
for both pip and easy_install).
Yes it's awful.
My $.02 is that if folks are concerned about their automated installs
breaking, they need to be using a private index or pinned versions, at
least until PyPI or a replacement grows features that make it not-insane
to use for production builds. The catalog-sig has an epic thread going
on about this right now. Embrace the suck.
- C
Sergey Schetinin
O_O
Right, so when someone gets such a hidden dev release then goes to
http://pypi.python.org/pypi/WebOb they will not even see it there.
That would be confusing :)
> My $.02 is that if folks are concerned about their automated installs
> breaking, they need to be using a private index or pinned versions, at
> least until PyPI or a replacement grows features that make it not-insane
> to use for production builds. The catalog-sig has an epic thread going
> on about this right now. Embrace the suck.
>
> - C
Joe Dallago
As far as Python 3 support goes, I already have a great deal of it done and plan on completing the rest this weekend in a development sprint so to speak. You can check out my latest blog post for details at jayd3e.com.
On Jul 7, 2011 7:34 PM, "Sergey Schetinin" <ser...@maluke.com> wrote:
On 8 July 2011 03:29, Chris McDonough <chr...@plope.com> wrote:
> On Fri, 2011-07-08 at 03:13 +0300,...
O_O
Right, so when someone gets such a hidden dev release then goes to
http://pypi.python.org/pypi/WebOb they will not even see it there.
That would be confusing :)
> My $.02 is that if folks are concerned about their automated installs
> breaking, they need to be...
http://self.maluke.com/
--
You received this message because you are subscribed to the Google Groups "Paste Users" group.
...
Alex Grönholm
> On 8 July 2011 03:29, Chris McDonough<chr...@plope.com> wrote:
>> On Fri, 2011-07-08 at 03:13 +0300, Sergey Schetinin wrote:
>>> 2011/7/8 Chris McDonough<chr...@plope.com>:
betas for established projects should not be published on PyPI at all.
It could break things for people who didn't think of specifying "webob
>= 1.0, < 1.1" or similar.
Distutils2/Packaging will make this a lot better in the future. Perhaps
you should also include packaging metadata in setup.cfg to facilitate
this? It certainly wouldn't hurt.
Chris McDonough
> 08.07.2011 03:34, Sergey Schetinin kirjoitti:
> > On 8 July 2011 03:29, Chris McDonough<chr...@plope.com> wrote:
> >> On Fri, 2011-07-08 at 03:13 +0300, Sergey Schetinin wrote:
> >>> 2011/7/8 Chris McDonough<chr...@plope.com>:
Yep. So be it is my opinion.
- C
Mike Orr
> The current state of the packaging tools is unfortunate, but alphas and
> betas for established projects should not be published on PyPI at all. It
> could break things for people who didn't think of specifying "webob >= 1.0,
> < 1.1" or similar.
Overspecifying 'webob<=1.0.99' leads to the opposite problem, that
packages will refuse to work together even when they're compatible.
You don't know until the next version is released whether it's going
to be compatible or even what version it will be. Over-restricting
leads to spurious exceptions which are hard to track down because the
error message doesn't tell which package is requiring a restricted
version. It becomes worse if the restriction is in a third-party
package that has become unmaintained or the maintainer has a different
philosophy than you, because then you have to modify a third party
package's setup.py to make it usable, and then you have to remember to
re-apply the modification at every deployment and every upgrade. So
there are dangers both ways in under-specifying and over-specifying
version restrictions, it's not just one way. And yes, I wish pip and
easy_install could be taught which kinds of packages are
stable/backward-compatible and which are not, rather than just relying
on the version number.
--
Mike Orr <slugg...@gmail.com>
Sergey Schetinin
http://pypi-stable.maluke.com/
It gets the data from pypi via xmlrpc interface and links to the
original downloads, but it filters out the alphas, betas and release
candidates. For ex. compare http://pypi.python.org/simple/WebOb/ and
http://pypi-stable.maluke.com/simple/WebOb/
The check if the version is final is simply this:
_rx_final = re.compile(r'[\d\.]+\Z')
def is_final(ver):
return bool(_rx_final.match(ver))
It can be used like this for ex.:
pip install -i http://pypi-stable.maluke.com/simple WebOb --upgrade
Look here for other ways to the non-canonical pypi indices:
http://jacobian.org/writing/when-pypi-goes-down/
There's some caching already built-in, but there's room for
improvement. It's hosted on GAE.
Let me know if something doesn't work, etc.
-Sergey
2011/7/8 Mike Orr <slugg...@gmail.com>:
> --
> You received this message because you are subscribed to the Google Groups "Paste Users" group.
> To post to this group, send email to paste...@googlegroups.com.
> To unsubscribe from this group, send email to paste-users...@googlegroups.com.
> For more options, visit this group at http://groups.google.com/group/paste-users?hl=en.
>
>
Ian Bicking
I guess the problem is really that the simple index is really chatty (http://pypi.python.org/simple/WebOb/) -- another option for betas might be a non-numeric version number, like beta.1.1.b1 -- I think that'll make it installable as WebOb==beta.1.1.b1 but won't make it the most recent version (ugly though it is).
Ian