Comment #57 on issue 2482 by tadej.ja...@gmail.com: Stop bundling mpmath
http://code.google.com/p/sympy/issues/detail?id=2482
P.S. Just discovered this post:
http://vagabond.github.io/2013/06/21/z_packagers-dont-know-best/ I am
posting it here to lighten up the discussion a bit :), but the
Comment #54 on issue 2482 by skirpic...@gmail.com: Stop bundling mpmath
http://code.google.com/p/sympy/issues/detail?id=2482
* more complex installation *everywhere*
What is a price? I don't think, that few words in README.rst to instruct
users first install mpmath actually poses a
Comment #55 on issue 2482 by asmeu...@gmail.com: Stop bundling mpmath
http://code.google.com/p/sympy/issues/detail?id=2482
It's already not enough: there is a lot of external dependencies in
sympy (e.g. gmpy).
There's a *big* difference here. SymPy works just fine without gmpy. You
just
Comment #56 on issue 2482 by skirpic...@gmail.com: Stop bundling mpmath
http://code.google.com/p/sympy/issues/detail?id=2482
There's a *big* difference here. SymPy works just fine without gmpy.
Yes, of course. But that's the next point (not related to harder
development and bug triaging).
Comment #53 on issue 2482 by ondrej.c...@gmail.com: Stop bundling mpmath
http://code.google.com/p/sympy/issues/detail?id=2482
P.S. Just discovered this post:
http://vagabond.github.io/2013/06/21/z_packagers-dont-know-best/ I am
posting it here to lighten up the discussion a bit :), but the
Comment #48 on issue 2482 by skirpic...@gmail.com: Stop bundling mpmath
http://code.google.com/p/sympy/issues/detail?id=2482
it does break with various earlier versions of mpmath
We can add workarounds, as for any other library.
So I think it should be quite easy to keep a copy of a working
Comment #49 on issue 2482 by asmeu...@gmail.com: Stop bundling mpmath
http://code.google.com/p/sympy/issues/detail?id=2482
What do you mean by mpmath is not a quite stable?
--
You received this message because this project is configured to send all
issue notifications to this address.
You
Comment #50 on issue 2482 by skirpic...@gmail.com: Stop bundling mpmath
http://code.google.com/p/sympy/issues/detail?id=2482
What do you mean by mpmath is not a quite stable?
http://code.google.com/p/sympy/issues/detail?id=2482#c30
--8--
First, I do not 100% agree with the statement that
Comment #51 on issue 2482 by fredrik@gmail.com: Stop bundling mpmath
http://code.google.com/p/sympy/issues/detail?id=2482
I can essentially guarantee that there will be no backwards incompatible
*interface* changes in mpmath.
As for bugs, it would obviously help to continue using
Comment #52 on issue 2482 by ondrej.c...@gmail.com: Stop bundling mpmath
http://code.google.com/p/sympy/issues/detail?id=2482
Arguments against unbundling:
* more complex installation *everywhere*
* harder development and bug triaging in sympy: suddenly sympy git hash is
not enough, you have
Comment #44 on issue 2482 by skirpic...@gmail.com: Stop bundling mpmath
http://code.google.com/p/sympy/issues/detail?id=2482
https://github.com/sympy/sympy/pull/2192
--
You received this message because this project is configured to send all
issue notifications to this address.
You may
Comment #45 on issue 2482 by ondrej.c...@gmail.com: Stop bundling mpmath
http://code.google.com/p/sympy/issues/detail?id=2482
Hi Tadej,
what you wrote also applies to Debian. I think we should make it so that if
people really want to depend on mpmath externally have the option to do so
---
Comment #46 on issue 2482 by asmeu...@gmail.com: Stop bundling mpmath
http://code.google.com/p/sympy/issues/detail?id=2482
Actually, we need to drop 2.5 to use relative imports, because import *
doesn't work with it due to some bug (if I remember correctly).
--
You received this message
Comment #47 on issue 2482 by tadej.ja...@gmail.com: Stop bundling mpmath
http://code.google.com/p/sympy/issues/detail?id=2482
Sergey,
thanks for stepping in and reviving this issue by preparing a pull request!
I like Ondřej's proposed approach. Indeed, this will simplify sympy's
maintenance
Comment #43 on issue 2482 by tadej.ja...@gmail.com: Stop bundling mpmath
http://code.google.com/p/sympy/issues/detail?id=2482
Hey!
I would just like to add one negative consequence of bundling mpmath.
Sympy has a Fedora package: https://apps.fedoraproject.org/packages/sympy,
but
Comment #41 on issue 2482 by ronan.l...@gmail.com: Stop bundling mpmath
http://code.google.com/p/sympy/issues/detail?id=2482
Relying on a hypothetical tarball that's updated once a year doesn't help
that much. It should be possible to install sympy on Python3 from a GitHub
clone.
--
You
Comment #42 on issue 2482 by asmeu...@gmail.com: Stop bundling mpmath
http://code.google.com/p/sympy/issues/detail?id=2482
It is.
./bin/use2to3
cd py3k-sympy
./setup.py install
You can even do setupegg.py develop, something that's not possible with the
setuptools method.
--
You received
Comment #38 on issue 2482 by brad.froehle: Stop bundling mpmath
http://code.google.com/p/sympy/issues/detail?id=2482
I've read through the comments and I appreciate those that deal with the
end user experience. However I think this points to an even stronger need
to unbundle mpmath. As an
Comment #39 on issue 2482 by brad.froehle: Stop bundling mpmath
http://code.google.com/p/sympy/issues/detail?id=2482
I've read through the comments and I appreciate those that deal with the
end user experience. However I think this points to an even stronger need
to unbundle mpmath. As an
Comment #40 on issue 2482 by vlada.pe...@gmail.com: Stop bundling mpmath
http://code.google.com/p/sympy/issues/detail?id=2482
Brad, the plan is for python3 setup.py install to just work, we'll just
be uploading separate tarballs with py2 and py3 sources (which has the
additional advantage
Updates:
Labels: -Priority-High Priority-Medium
Comment #36 on issue 2482 by ondrej.c...@gmail.com: Stop bundling mpmath
http://code.google.com/p/sympy/issues/detail?id=2482
I am lowering the priority. The py3 thing is fixed, and I am -1 on removing
mpmath.
--
You received this
Comment #37 on issue 2482 by ronan.l...@gmail.com: Stop bundling mpmath
http://code.google.com/p/sympy/issues/detail?id=2482
The py3 thing is only sort-of fixed, because now we have to deal with bugs
in our toolchain, like issue 2890.
--
You received this message because you are subscribed
Updates:
Labels: -Milestone-Release0.7.2
Comment #35 on issue 2482 by vlada.pe...@gmail.com: Stop bundling mpmath
http://code.google.com/p/sympy/issues/detail?id=2482
Well, in the end, we did the bin/use2to3 thing and it works well enough. I
still think mpmath should be removed, but
Issue 2482: Stop bundling mpmath
http://code.google.com/p/sympy/issues/detail?id=2482
This issue is no longer blocking issue 1262.
See http://code.google.com/p/sympy/issues/detail?id=1262
--
You received this message because you are listed in the owner
or CC fields of this issue, or because you
Comment #33 on issue 2482 by vlada.pe...@gmail.com: Stop bundling mpmath
http://code.google.com/p/sympy/issues/detail?id=2482
Some minor points:
* py.test: I was under the impression that our testing framework is a fork
of an old version of py.test. Anyway, the argument is that if we had
Comment #25 on issue 2482 by asmeurer: Stop bundling mpmath
http://code.google.com/p/sympy/issues/detail?id=2482
I just realized that porting to Distribute itself would create an install
dependency, as it is not currently bundled with Python by default.
Vladimir, you've looked into this.
Comment #26 on issue 2482 by asmeurer: Stop bundling mpmath
http://code.google.com/p/sympy/issues/detail?id=2482
OK, I found your question on StackOverflow
http://stackoverflow.com/questions/6344076/differences-between-distribute-distutils-and-setuptools,
which answers my question. It
Comment #27 on issue 2482 by ondrej.c...@gmail.com: Stop bundling mpmath
http://code.google.com/p/sympy/issues/detail?id=2482
I think we all have said our arguments, I read through the discussion
again, and I think that for pyglet, all Vlads+Ronan's arguments apply, and
it is now gone.
Comment #28 on issue 2482 by vlada.pe...@gmail.com: Stop bundling mpmath
http://code.google.com/p/sympy/issues/detail?id=2482
The 2to3 issue: as mpmath code is compatible with both py2 and py3 with the
same code (no 2to3 involved) calling 2to3 on it produces bad code. One
banal example is:
Comment #29 on issue 2482 by ondrej.c...@gmail.com: Stop bundling mpmath
http://code.google.com/p/sympy/issues/detail?id=2482
2to3 issue: that's pretty bad. Obviously, one solution is to unbundle
mpmath, I understand that.
Keeping it in, the only way is to make sure 2to3 is not run on it. I
Comment #30 on issue 2482 by asmeurer: Stop bundling mpmath
http://code.google.com/p/sympy/issues/detail?id=2482
I think we all have said our arguments
I actually disagree with this statement. A lot of core members of the
community have not yet chimed in on this, and I would like to hear
Comment #31 on issue 2482 by ondrej.c...@gmail.com: Stop bundling mpmath
http://code.google.com/p/sympy/issues/detail?id=2482
Ah, I am -1 to use distribute, I forgot to write it. We should stick with
what is in pure Python. The 2to3 tool seems to have quite a few problems
apparently, if it
Comment #32 on issue 2482 by asmeurer: Stop bundling mpmath
http://code.google.com/p/sympy/issues/detail?id=2482
Ah, that is something I forgot about. We can (and should) just provide the
tarballs for 3.0 separately from 2.5-7, so that the user never has to run
2to3 unless he's developing
Comment #23 on issue 2482 by ronan.l...@gmail.com: Stop bundling mpmath
http://code.google.com/p/sympy/issues/detail?id=2482
After discussing with Vladimir, it seems that it's basically impossible to
hack Distribute to use a custom version of 2to3. The problem is that it
already implements
Comment #22 on issue 2482 by vlada.pe...@gmail.com: Stop bundling mpmath
http://code.google.com/p/sympy/issues/detail?id=2482
First of all, why would mpmath become incompatible? It's a library like any
other, it should first deprecate functions for a few releases and only then
remove them
Comment #16 on issue 2482 by ronan.l...@gmail.com: Stop bundling mpmath
http://code.google.com/p/sympy/issues/detail?id=2482
sympy is broken for some users now, because we don't handle dependencies
properly. Just ask any Mac user about plotting.
Could describe your use case precisely? I'm
Comment #17 on issue 2482 by ondrej.c...@gmail.com: Stop bundling mpmath
http://code.google.com/p/sympy/issues/detail?id=2482
Let's keep this issue focused to *mpmath* only.
(For pyglet, I totally agree with you, that it is broken, and I agree with
you, that we should not ship pyglet with
Comment #18 on issue 2482 by vlada.pe...@gmail.com: Stop bundling mpmath
http://code.google.com/p/sympy/issues/detail?id=2482
Well, Firefox cannot run without xulrunner, but they don't come together.
And they're even developed by the same people!
--
You received this message because you are
Comment #19 on issue 2482 by ondrej.c...@gmail.com: Stop bundling mpmath
http://code.google.com/p/sympy/issues/detail?id=2482
When I install firefox, I don't install xulrunner. I download the binary
and it just works. I am pretty sure, that most people that use firefox
either use it from
Comment #20 on issue 2482 by ronan.l...@gmail.com: Stop bundling mpmath
http://code.google.com/p/sympy/issues/detail?id=2482
If we don't bundle mpmath, pip install sympy will still work, sudo apt-get
install python-sympy will still work (and
Comment #21 on issue 2482 by ondrej.c...@gmail.com: Stop bundling mpmath
http://code.google.com/p/sympy/issues/detail?id=2482
Thanks for discussing it. Assume that new mpmath version stops being
compatible. What will you do? I assume that you will update sympy and
release as well, right
Comment #9 on issue 2482 by fredrik@gmail.com: Stop bundling mpmath
http://code.google.com/p/sympy/issues/detail?id=2482
Yes, mpmath can be considered stable (and has been for a few versions).
There should be no future changes that affect compatibility with SymPy.
--
You received this
Comment #10 on issue 2482 by ondrej.c...@gmail.com: Stop bundling mpmath
http://code.google.com/p/sympy/issues/detail?id=2482
I am -1 on this. On linux there is no problem installing all the
dependencies, on Mac it is a bit harder (no apt-get install scipy), and on
Windows it is way harder.
Comment #11 on issue 2482 by ondrej.c...@gmail.com: Stop bundling mpmath
http://code.google.com/p/sympy/issues/detail?id=2482
Btw, it is actually a problem on linux as well --- for example at work, I
don't have root access to my computer.
The bigger solution is for example EPD, Sage or
Comment #12 on issue 2482 by fredrik@gmail.com: Stop bundling mpmath
http://code.google.com/p/sympy/issues/detail?id=2482
A possibility would be to remove sympy.mpmath but still bundle mpmath with
sympy (there could be a sympy download without mpmath, for use by distros),
assuming that
Comment #13 on issue 2482 by vlada.pe...@gmail.com: Stop bundling mpmath
http://code.google.com/p/sympy/issues/detail?id=2482
First of all, I have to say that I don't consider ease of porting to Py3 a
major argument. Yes, it makes things a little easier, but as Aaron said, I
can work around
Comment #14 on issue 2482 by ronan.l...@gmail.com: Stop bundling mpmath
http://code.google.com/p/sympy/issues/detail?id=2482
There's another major reason for not bundling external libraries:
avoiding DLL hell, i.e. the situation where there are multiple versions
of the same library
Comment #15 on issue 2482 by ondrej.c...@gmail.com: Stop bundling mpmath
http://code.google.com/p/sympy/issues/detail?id=2482
There are several packages of mpmath in Debian, and some of the older ones
do break sympy. When I, as a user, download sympy, I expect it to just
work. If I want to
Status: NeedsDecision
Owner: ronan.l...@gmail.com
CC: vlada.pe...@gmail.com
Labels: Type-Defect Priority-High Milestone-Release0.7.1
New issue 2482 by ronan.l...@gmail.com: Stop bundling mpmath
http://code.google.com/p/sympy/issues/detail?id=2482
Bundling mpmath causes problems for
Issue 2482: Stop bundling mpmath
http://code.google.com/p/sympy/issues/detail?id=2482
This issue is now blocking issue 1262.
See http://code.google.com/p/sympy/issues/detail?id=1262
--
You received this message because you are listed in the owner
or CC fields of this issue, or because you
Updates:
Cc: ondrej.c...@gmail.com
Comment #2 on issue 2482 by asmeurer: Stop bundling mpmath
http://code.google.com/p/sympy/issues/detail?id=2482
Ondrej, could you weigh in on this?
--
You received this message because you are subscribed to the Google Groups
sympy-issues group.
To
Comment #3 on issue 2482 by asmeurer: Stop bundling mpmath
http://code.google.com/p/sympy/issues/detail?id=2482
Regarding problems with 2to3, I was telling Vladimir that we might want to
just write our own ./bin/2to3 tool. This could have some advantages:
- You wouldn't have to add all the
Comment #4 on issue 2482 by ronan.l...@gmail.com: Stop bundling mpmath
http://code.google.com/p/sympy/issues/detail?id=2482
Obviously, 2to3 is just a wrapper for lib2to3.main(). We'd have to fork
that function in order to create our own 2to3. And then we'd have to fork
the install command
Updates:
Cc: fredrik@gmail.com Vinzent.Steinberg
Comment #5 on issue 2482 by asmeurer: Stop bundling mpmath
http://code.google.com/p/sympy/issues/detail?id=2482
Why would we have to fork lib2to3.main()? Just write a wrapper around it
(that calls it) that automatically gives the
Comment #6 on issue 2482 by asmeurer: Stop bundling mpmath
http://code.google.com/p/sympy/issues/detail?id=2482
By the way, I didn't know about Distribute's use_2to3.
Honestly, these tools should support being able to skip files. Maybe we
should backport some improvements.
--
You received
Comment #7 on issue 2482 by ronan.l...@gmail.com: Stop bundling mpmath
http://code.google.com/p/sympy/issues/detail?id=2482
Ah yes, passing the list of all the files and directories below sympy/
should work. That won't help with Distribute though.
--
You received this message because you
Comment #8 on issue 2482 by asmeurer: Stop bundling mpmath
http://code.google.com/p/sympy/issues/detail?id=2482
Ronan, would you consider it to be within the scope of Vladimir's project
to backport improvements to other projects that would make his life easier,
like making 2to3/Distribute
57 matches
Mail list logo