matplotlib-devel Mailing List for matplotlib

On 10/07/2014 05:13 PM, Eric Firing wrote:
> I'm not sure what to do about
> https://github.com/matplotlib/matplotlib/issues/3621 either. In
> general, I think it is best to have strings like this left in their
> native mode, not coerced to unicode, but that is contrary to the
> strategy decided upon when "from __future__ import unicode_literals" was
> adopted.
I think it has a pretty easy fix -- just put `str()` around the version 
string. That will go back to bytes on 2.x and unicode on 3.x.
>
> Mike, what do you think--are other projects going to run into this
> problem, triggered by a bug in LooseVersion?
It's an unfortunate bug, but fortunately has a reasonable workaround.
I think there are certain areas of matplotlib where the benefit of using 
`unicode_literals` outweighs the pain -- particularly in the text and 
math text handling, the PDF and SVG backends, and anything to do with 
fonts. It's actually fixed a lot of bugs for us. But, yes, there are 
the occasional dark corners in Python 2 like this.
Mike
>
> Eric
>
> On 2014年10月07日, 11:03 AM, Michael Droettboom wrote:
>> Thanks for working on all of these. I just discovered another issue
>> that really should be a blocker:
>>
>> https://github.com/matplotlib/matplotlib/issues/3622
>>
>> Mike
>>
>> On 10/01/2014 11:58 PM, Thomas Caswell wrote:
>>> Hello all,
>>>
>>> We are going to miss the deadline on 1.4.1 as there is 2-3 blocker issues:
>>>
>>> - #3470 / PR#3564 which started as issues with the macosx backend
>>> and spiralled into discovering that we were only validating input to
>>> rcparams about half of the time.
>>>
>>> - #3505 The changes to disable interactive mode when not at a repl.
>>> It turns out a lot of people use it and we should un-break them.
>>>
>>> - #3517 which is related to non-ascii paths in font look up which
>>> causes matplotlib to blow up on import.
>>>
>>> I am open to arguments that any of these should not be blockers.
>>>
>>> Tom
>>>
>>
>
> ------------------------------------------------------------------------------
> Meet PCI DSS 3.0 Compliance Requirements with EventLog Analyzer
> Achieve PCI DSS 3.0 Compliant Status with Out-of-the-box PCI DSS Reports
> Are you Audit-Ready for PCI DSS 3.0 Compliance? Download White paper
> Comply to PCI DSS 3.0 Requirement 10 and 11.5 with EventLog Analyzer
> http://pubads.g.doubleclick.net/gampad/clk?id=154622311&iu=/4140/ostg.clktrk
> _______________________________________________
> Matplotlib-devel mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-devel
-- 
Michael Droettboom
Science Software Branch
Space Telescope Science Institute
http://www.droettboom.com

I'm not sure what to do about 
https://github.com/matplotlib/matplotlib/issues/3621 either. In 
general, I think it is best to have strings like this left in their 
native mode, not coerced to unicode, but that is contrary to the 
strategy decided upon when "from __future__ import unicode_literals" was 
adopted.
Mike, what do you think--are other projects going to run into this 
problem, triggered by a bug in LooseVersion?
Eric
On 2014年10月07日, 11:03 AM, Michael Droettboom wrote:
> Thanks for working on all of these. I just discovered another issue
> that really should be a blocker:
>
> https://github.com/matplotlib/matplotlib/issues/3622
>
> Mike
>
> On 10/01/2014 11:58 PM, Thomas Caswell wrote:
>> Hello all,
>>
>> We are going to miss the deadline on 1.4.1 as there is 2-3 blocker issues:
>>
>> - #3470 / PR#3564 which started as issues with the macosx backend
>> and spiralled into discovering that we were only validating input to
>> rcparams about half of the time.
>>
>> - #3505 The changes to disable interactive mode when not at a repl.
>> It turns out a lot of people use it and we should un-break them.
>>
>> - #3517 which is related to non-ascii paths in font look up which
>> causes matplotlib to blow up on import.
>>
>> I am open to arguments that any of these should not be blockers.
>>
>> Tom
>>
>
>

Thanks for working on all of these. I just discovered another issue 
that really should be a blocker:
https://github.com/matplotlib/matplotlib/issues/3622
Mike
On 10/01/2014 11:58 PM, Thomas Caswell wrote:
> Hello all,
>
> We are going to miss the deadline on 1.4.1 as there is 2-3 blocker issues:
>
> - #3470 / PR#3564 which started as issues with the macosx backend
> and spiralled into discovering that we were only validating input to
> rcparams about half of the time.
>
> - #3505 The changes to disable interactive mode when not at a repl.
> It turns out a lot of people use it and we should un-break them.
>
> - #3517 which is related to non-ascii paths in font look up which
> causes matplotlib to blow up on import.
>
> I am open to arguments that any of these should not be blockers.
>
> Tom
>
-- 
Michael Droettboom
Science Software Branch
Space Telescope Science Institute
http://www.droettboom.com