Python 2.5 ; Effbot console ; thank ; pb release.

[Méta-MCI]

Hi! (***sorry for my approximative english***)


A few months ago, I needed a console, under Windows.
After several research, I selected the console of EffBot.

Thank you very much, Fredrik Lundh, for this small tool,
quite practical and which repaired me well.


Then, Python 2.5 arrived.

Many modules/libraries was adapted to Python 2.5

But, it missed the console of EffBot.

Patiently, I visited the site of EffBot regularly.
Then, I started to corrode me the nails.
Then, I questioned myself:
"with each new version of Python, certain libraries are never
adapted/compiled."

For the professional developments, it is a major risk.
For (the evolution of) Python, it is an obstacle.

The rebuilding of libraries (not-standard) is not (under Windows) very easy,
and not very standard.
To facilitate this possibility (of rebuilding) could be an objective,
intended to improve perenniality of Python.

What do you think of that?



(
While waiting, for the console, I ended up writing my (perso) small
module.

Therefore, THANK YOU, Fredrik, FOR NOR TO HAVE COMPILED a version for
P-2.5,
because:
- that made me become aware of brittleness, in time, of the external
bookshops;
- that forced me to write a trick of which I will be used again
myself.
)


@-salutations

Michel Claveau

 

[Fredrik Lundh]

For the professional developments, it is a major risk.

some days, I ask myself why I shouldn't just use GPL for everything I
do, and ship it as source code only.

</F>

 

[Steve Holden]

Méta-MCI wrote:




some days, I ask myself why I shouldn't just use GPL for everything I
do, and ship it as source code only.

To which I presume the answer is that you are considerate of Windows
users who'd rather not compile their own Windows applications due to the
cost of using the commercial compilers and the complexity of using the
open source ones.

Whatever the reasons, a vote of thanks to all extension authors who *do*
bother to compile for Windows (despite complaints from others who don't
feel this happens quickly enough).

regards
Steve

 

[Paul Boddie]

I'll cut in here and mention that it's a risk that can be managed
through various well understood methods of deployment. For me, Python
2.4 is going to be good enough until (and even beyond) the time I can
be bothered to either upgrade my distribution's packages or to upgrade
my distribution and get the packages for Python 2.5. At which point,
I'll just need to select the pinnacles of my package tower; then I can
hopefully just press the button and have everything working on Python
2.5.

Generally, package developers shouldn't be "aggressively" using the
absolute latest and greatest version's features, and if you're pursuing
"professional developments" you might want to keep using the mature
releases of Python whilst letting other developers know that adoption
of their stuff will be limited if they make it part of a rapidly moving
target. Sure, I can imagine that people are desperate to use the "with"
statement all over the place, but maintainers of widely used (or
"professional") stuff have to exercise some conservatism - otherwise
someone has to start forking and backporting their changes.

To which I presume the answer is that you are considerate of Windows
users who'd rather not compile their own Windows applications due to the
cost of using the commercial compilers and the complexity of using the
open source ones.

Well, there's a commercial service you could offer, Steve. ;-)

Whatever the reasons, a vote of thanks to all extension authors who *do*
bother to compile for Windows (despite complaints from others who don't
feel this happens quickly enough).

This happens every time a new release of Python comes out: people want
to play with the new features, but they experience a period of
frustration because their favourite packages aren't available. I'd
advise people to download the installer, get their fill of the new
features for a few minutes, then schedule another look in a few weeks.
Or they can start paying people to make it all happen "yesterday", of
course.

Paul

 

[Steve Holden]

I'll cut in here and mention that it's a risk that can be managed
through various well understood methods of deployment. For me, Python
2.4 is going to be good enough until (and even beyond) the time I can
be bothered to either upgrade my distribution's packages or to upgrade
my distribution and get the packages for Python 2.5. At which point,
I'll just need to select the pinnacles of my package tower; then I can
hopefully just press the button and have everything working on Python
2.5.

Generally, package developers shouldn't be "aggressively" using the
absolute latest and greatest version's features, and if you're pursuing
"professional developments" you might want to keep using the mature
releases of Python whilst letting other developers know that adoption
of their stuff will be limited if they make it part of a rapidly moving
target. Sure, I can imagine that people are desperate to use the "with"
statement all over the place, but maintainers of widely used (or
"professional") stuff have to exercise some conservatism - otherwise
someone has to start forking and backporting their changes.




Well, there's a commercial service you could offer, Steve. ;-)
You think people would pay?



This happens every time a new release of Python comes out: people want
to play with the new features, but they experience a period of
frustration because their favourite packages aren't available. I'd
advise people to download the installer, get their fill of the new
features for a few minutes, then schedule another look in a few weeks.
Or they can start paying people to make it all happen "yesterday", of
course.

If there's evidence of demand this *is* something I'd consider doing.

regards
Steve

 

[Ben Finney]

some days, I ask myself why I shouldn't just use GPL for everything
I do, and ship it as source code only.

Many popular programs *do* ship as source only, and leave it to the
community to build and make available binaries for platforms that the
community wants. This way, the developers don't have to maintain
builds for anything but the platforms they actually care about, and a
particular platform is supported only to the extent that third parties
put effort into it and share the result.

I don't see why you shouldn't do the same.

 

[Hendrik van Rooyen]

For the professional developments, it is a major risk.

some days, I ask myself why I shouldn't just use GPL for everything I
do, and ship it as source code only.

</F>

because then you would deny your work to thousands of ungrateful, unmotivated
lazy buggers like me...

seriously, though- the OP's problem was that he was fixing something that was
not broken - by upgrading, for some incomprehensible reason - and then
expecting that the whole world would follow, and ease his pain by intuiting what
he was using, and upgrading it for him, to make his chosen add ons compatible
with his upgrade...
I wonder what else his upgrade broke? - you obviously have a major task on your
hands as you are expected to fix that too...

Why are you surprised at ingratitude, BTW?

Keep up the good work.

- Hendrik

 

[Ben Finney]

[quoting problems fixed]

because then you would deny your work to thousands of ungrateful,
unmotivated lazy buggers like me...

Not necessarily. All it needs is one person (with the same platform
you want to use) to take the source, build it for that platform, and
make it available. All the other "ungrateful, unmotivated lazy
buggers" can then take advantage of that -- and reward the person with
whatever praise they require

 

[Thomas Heller]

[quoting problems fixed]

because then you would deny your work to thousands of ungrateful,
unmotivated lazy buggers like me...

Not necessarily. All it needs is one person (with the same platform
you want to use) to take the source, build it for that platform, and
make it available. All the other "ungrateful, unmotivated lazy
buggers" can then take advantage of that -- and reward the person with
whatever praise they require

I wonder if it would be possible to setup a windows box which provides
a (web-)service that allows to build Python packages.

Any ideas how this could be made secure?

Thomas

 

[Laurent Pointal]

Thomas Heller a écrit :

[quoting problems fixed]

some days, I ask myself why I shouldn't just use GPL for
everything I do, and ship it as source code only.
because then you would deny your work to thousands of ungrateful,
unmotivated lazy buggers like me...

Not necessarily. All it needs is one person (with the same platform
you want to use) to take the source, build it for that platform, and
make it available. All the other "ungrateful, unmotivated lazy
buggers" can then take advantage of that -- and reward the person with
whatever praise they require

I wonder if it would be possible to setup a windows box which provides
a (web-)service that allows to build Python packages.

You will have the problem of third-party libraries dependancies (and
installation).

 

[Paul Boddie]

I wonder if it would be possible to setup a windows box which provides
a (web-)service that allows to build Python packages.

Isn't this the kind of service your distribution vendor should be
providing, especially if you've paid them good money? Oh wait! That
would be Microsoft, wouldn't it? ;-)

Still, there's always ActiveState, and I guess they still have that PPM
installer utility for people wanting some kind of binary package
download mechanism.

Paul

 

[=?ISO-8859-1?Q?=22Martin_v=2E_L=F6wis=22?=]

Not necessarily. All it needs is one person (with the same platform
you want to use) to take the source, build it for that platform, and
make it available.

In my experience, this single person *never* exists. I.e. it's either
the original author who provides binaries, or none are provided.
People certainly build it themselves, but then never share. I don't
know why that is - probably, it takes a different mindset to be on
the giving end than on the receiving end of free software.

There are exceptions, of course: when communities are built. There
is no question that binaries are available with the Linux distributions,
and the various Sun, HP, IBM, Darwinports, etc. free software sites also
have an established contributorship. For Windows, such communities
apparently don't form, for some reason.

Regards,
Martin

 

[Istvan Albert]

People certainly build it themselves, but then never share. I don't
know why that is - probably, it takes a different mindset to be on
the giving end than on the receiving end of free software.

I've seen people share binary builds, for example you can get the
windows psycopg from a third party

I think most people may not be aware that there is a need for such a
build, or have logistical issues (where to post it), and maybe assume
that sooner rather than later the original author will have one such
build anyhow and that makes their own effort (however minute)
pointless.

i.

 

[Ben Finney]

In my experience [...] it's either the original author who provides
binaries, or none are provided. People certainly build it
themselves, but then never share.

IIRC, the early days of Mozilla (after the code was released as free
software) had the developers releasing source code, and the community
providing binary builds.

Battle for Wesnoth also works this way.

There are exceptions, of course: when communities are built. There
is no question that binaries are available with the Linux
distributions, and the various Sun, HP, IBM, Darwinports, etc. free
software sites also have an established contributorship. For
Windows, such communities apparently don't form, for some reason.

All the cases I can recall are for projects where the code is designed
to work on multiple architectures without much tweaking. So it seems
to fit this case fine.

 

[Magnus Lycka]

Hi! (***sorry for my approximative english***)

That's ok. Quite amusing to read that you were repaired.

A few months ago, I needed a console, under Windows.
After several research, I selected the console of EffBot.

Thank you very much, Fredrik Lundh, for this small tool,
quite practical and which repaired me well.

Then, Python 2.5 arrived.

That doesn't mean it's clever to use 2.5 right now...
There's plenty of code which isn't adapted to 2.5 yet.

Personally, I use 2.4 at work (although sometimes I
still need to be 2.2 compatible) and at home I have
both 2.4 and 2.5 installed. Most of the time I work
with 2.4 since I use packages which don't support 2.5
yet (or at least the last time I checked).