[Matplotlib-devel] v2.0.0b2

classic Classic list List threaded Threaded
7 messages Options
Reply | Threaded
Open this post in threaded view
|

[Matplotlib-devel] v2.0.0b2

tcaswell
Folks,

Last night (and in time for the first day of scipy 2016) we tagged v2.0.0b2.

Hopefully this will be the last beta, and we will have an rc soon after scipy.

By the end of the day I would like to have a (source) pre release on pypi and available via conda forge.

Tom

_______________________________________________
Matplotlib-devel mailing list
[hidden email]
https://mail.python.org/mailman/listinfo/matplotlib-devel
Reply | Threaded
Open this post in threaded view
|

Re: v2.0.0b2

matplotlib - devel mailing list
Did you take a look at the image I posted regarding the font issue I mentioned with regards tables?



From: Thomas Caswell <[hidden email]>
To: matplotlib development list <[hidden email]>
Sent: Wednesday, 13 July 2016, 20:48
Subject: [Matplotlib-devel] v2.0.0b2

Folks,

Last night (and in time for the first day of scipy 2016) we tagged v2.0.0b2.

Hopefully this will be the last beta, and we will have an rc soon after scipy.

By the end of the day I would like to have a (source) pre release on pypi and available via conda forge.

Tom

_______________________________________________
Matplotlib-devel mailing list
[hidden email]
https://mail.python.org/mailman/listinfo/matplotlib-devel



_______________________________________________
Matplotlib-devel mailing list
[hidden email]
https://mail.python.org/mailman/listinfo/matplotlib-devel
Reply | Threaded
Open this post in threaded view
|

Re: v2.0.0b2

Matthew Brett
In reply to this post by tcaswell
Hi,

On Wed, Jul 13, 2016 at 7:48 PM, Thomas Caswell <[hidden email]> wrote:
> Folks,
>
> Last night (and in time for the first day of scipy 2016) we tagged v2.0.0b2.
>
> Hopefully this will be the last beta, and we will have an rc soon after
> scipy.
>
> By the end of the day I would like to have a (source) pre release on pypi
> and available via conda forge.

Wheels building over at
https://travis-ci.org/MacPython/matplotlib-wheels/builds/144560516

32-bit wheels failing as before.

OSX builds delayed because of a travis-ci OSX backlog.

OK to upload when done?

Cheers,

Matthew
_______________________________________________
Matplotlib-devel mailing list
[hidden email]
https://mail.python.org/mailman/listinfo/matplotlib-devel
Reply | Threaded
Open this post in threaded view
|

Re: v2.0.0b2

tcaswell
@Matthew Yes please, I just push the source tarball up and am working on sorting out how to turn the crank on the many linux docker

@oceanwolf yes, but I did not understand the issue.

On Wed, Jul 13, 2016 at 4:59 PM Matthew Brett <[hidden email]> wrote:
Hi,

On Wed, Jul 13, 2016 at 7:48 PM, Thomas Caswell <[hidden email]> wrote:
> Folks,
>
> Last night (and in time for the first day of scipy 2016) we tagged v2.0.0b2.
>
> Hopefully this will be the last beta, and we will have an rc soon after
> scipy.
>
> By the end of the day I would like to have a (source) pre release on pypi
> and available via conda forge.

Wheels building over at
https://travis-ci.org/MacPython/matplotlib-wheels/builds/144560516

32-bit wheels failing as before.

OSX builds delayed because of a travis-ci OSX backlog.

OK to upload when done?

Cheers,

Matthew

_______________________________________________
Matplotlib-devel mailing list
[hidden email]
https://mail.python.org/mailman/listinfo/matplotlib-devel
Reply | Threaded
Open this post in threaded view
|

Re: v2.0.0b2

matplotlib - devel mailing list
Okay, I have cropped and enlarged the image (see attached) for you to see the problem with how the example renders on my computer.  As you see all of "y"s in "year" sit flush with the bottom (no gap/padding) apart from "20 year" which has a 1px gap from the bottom of the "y" to edge of the cell.  Notice also that the "0"s on 20 and 10 have a one pixel gap above them, whilst on the others they have a 2px gap.

I don't know where the trouble comes from, I haven't looked into the text part of the codebase yet, so I don't know how we calculate the row height and whether/how we use external font-libraries to do some of this for us, so ignoring that, I would expect it to (at least in this simple table situation) calculate a theoretical max height, i.e. from the bottom of the glyph with the lowest descender to the top of the glyph with the highest ascender, and then add whatever padding we specify around that and use that for every text row, so that the imaginary "ascender height" and "descender height" lines seen in the figure here <a id="yui_3_16_0_1_1468409528391_53563" rel="nofollow" shape="rect" target="_blank" href="javascript:void(0)">https://en.wikipedia.org/wiki/Ascender_%28typography%29 appear in the same place in each cell.  I have oversimplified my expected logic here for sensible default scenario and assuming 1 line of text per cell, and nothing other than text in the table.

Does that clarify the problem?  I thought that as we touch the default style here we might as well get it looking perfect, but if I go too nitpicky then we can save it for a bug-fix release / feature release.

Best,
OceanWolf



From: Thomas Caswell <[hidden email]>
To: Matthew Brett <[hidden email]>
Cc: matplotlib development list <[hidden email]>
Sent: Thursday, 14 July 2016, 0:23
Subject: Re: [Matplotlib-devel] v2.0.0b2

@Matthew Yes please, I just push the source tarball up and am working on sorting out how to turn the crank on the many linux docker

@oceanwolf yes, but I did not understand the issue.

_______________________________________________
Matplotlib-devel mailing list
[hidden email]
https://mail.python.org/mailman/listinfo/matplotlib-devel

table_new_zoomed.png (13K) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: v2.0.0b2

Benjamin Root
might be an even/odd thing?

On Wed, Jul 13, 2016 at 7:26 PM, OceanWolf via Matplotlib-devel <[hidden email]> wrote:
Okay, I have cropped and enlarged the image (see attached) for you to see the problem with how the example renders on my computer.  As you see all of "y"s in "year" sit flush with the bottom (no gap/padding) apart from "20 year" which has a 1px gap from the bottom of the "y" to edge of the cell.  Notice also that the "0"s on 20 and 10 have a one pixel gap above them, whilst on the others they have a 2px gap.

I don't know where the trouble comes from, I haven't looked into the text part of the codebase yet, so I don't know how we calculate the row height and whether/how we use external font-libraries to do some of this for us, so ignoring that, I would expect it to (at least in this simple table situation) calculate a theoretical max height, i.e. from the bottom of the glyph with the lowest descender to the top of the glyph with the highest ascender, and then add whatever padding we specify around that and use that for every text row, so that the imaginary "ascender height" and "descender height" lines seen in the figure here https://en.wikipedia.org/wiki/Ascender_%28typography%29 appear in the same place in each cell.  I have oversimplified my expected logic here for sensible default scenario and assuming 1 line of text per cell, and nothing other than text in the table.

Does that clarify the problem?  I thought that as we touch the default style here we might as well get it looking perfect, but if I go too nitpicky then we can save it for a bug-fix release / feature release.

Best,
OceanWolf



From: Thomas Caswell <[hidden email]>
To: Matthew Brett <[hidden email]>
Cc: matplotlib development list <[hidden email]>
Sent: Thursday, 14 July 2016, 0:23
Subject: Re: [Matplotlib-devel] v2.0.0b2

@Matthew Yes please, I just push the source tarball up and am working on sorting out how to turn the crank on the many linux docker

@oceanwolf yes, but I did not understand the issue.

_______________________________________________
Matplotlib-devel mailing list
[hidden email]
https://mail.python.org/mailman/listinfo/matplotlib-devel



_______________________________________________
Matplotlib-devel mailing list
[hidden email]
https://mail.python.org/mailman/listinfo/matplotlib-devel
Reply | Threaded
Open this post in threaded view
|

Re: v2.0.0b2

Ryan May-3
Maybe related to some kind of fractional pixel shift? Every so often, it accumulates enough to shift a whole pixel?

Ryan

On Wed, Jul 13, 2016 at 10:29 PM, Benjamin Root <[hidden email]> wrote:
might be an even/odd thing?

On Wed, Jul 13, 2016 at 7:26 PM, OceanWolf via Matplotlib-devel <[hidden email]> wrote:
Okay, I have cropped and enlarged the image (see attached) for you to see the problem with how the example renders on my computer.  As you see all of "y"s in "year" sit flush with the bottom (no gap/padding) apart from "20 year" which has a 1px gap from the bottom of the "y" to edge of the cell.  Notice also that the "0"s on 20 and 10 have a one pixel gap above them, whilst on the others they have a 2px gap.

I don't know where the trouble comes from, I haven't looked into the text part of the codebase yet, so I don't know how we calculate the row height and whether/how we use external font-libraries to do some of this for us, so ignoring that, I would expect it to (at least in this simple table situation) calculate a theoretical max height, i.e. from the bottom of the glyph with the lowest descender to the top of the glyph with the highest ascender, and then add whatever padding we specify around that and use that for every text row, so that the imaginary "ascender height" and "descender height" lines seen in the figure here https://en.wikipedia.org/wiki/Ascender_%28typography%29 appear in the same place in each cell.  I have oversimplified my expected logic here for sensible default scenario and assuming 1 line of text per cell, and nothing other than text in the table.

Does that clarify the problem?  I thought that as we touch the default style here we might as well get it looking perfect, but if I go too nitpicky then we can save it for a bug-fix release / feature release.

Best,
OceanWolf



From: Thomas Caswell <[hidden email]>
To: Matthew Brett <[hidden email]>
Cc: matplotlib development list <[hidden email]>
Sent: Thursday, 14 July 2016, 0:23
Subject: Re: [Matplotlib-devel] v2.0.0b2

@Matthew Yes please, I just push the source tarball up and am working on sorting out how to turn the crank on the many linux docker

@oceanwolf yes, but I did not understand the issue.

_______________________________________________
Matplotlib-devel mailing list
[hidden email]
https://mail.python.org/mailman/listinfo/matplotlib-devel



_______________________________________________
Matplotlib-devel mailing list
[hidden email]
https://mail.python.org/mailman/listinfo/matplotlib-devel




--
Ryan May


_______________________________________________
Matplotlib-devel mailing list
[hidden email]
https://mail.python.org/mailman/listinfo/matplotlib-devel