Mailman 3 September 2012 - scikit-image

Re: Numpydoc question
by Tony Yu Sept. 2, 2012

Sept. 2, 2012

On Sun, Sep 2, 2012 at 5:05 PM, Stéfan van der Walt <stefan(a)sun.ac.za>wrote: > On Sep 2, 2012 7:15 PM, "Tony Yu" <tsyu80(a)gmail.com> wrote: > > > > While looking through the sphinx config file, I noticed that we have > config settings for the `plot_directive` extension and also include the > sphinx extension in `doc/ext/`. A quick grep of the doc directory suggests > that we don't actually use the plot directive anywhere. Is it worth keeping > for future use or should I remove that as part of this doc clean up? > > There's probably some legacy cruft left over--few people ever read that > file, so feel free to clean it up. > > Is the plot directive non needed to build figures for the user guide? > > Thanks! > Stéfan > > > > -Tony > It's not needed for the user guide, but I later noticed that it's used in a couple of the docstrings (e.g. API docs of hough transform). I'll just leave the plot directive as is, since it is actually being used. I did manage to fix a number of Sphinx warnings. See PR 295<https://github.com/scikits-image/scikits-image/pull/295/files> . -Tony

1 0

Re: Numpydoc question
by Stéfan van der Walt Sept. 2, 2012

Sept. 2, 2012

On Sep 2, 2012 7:15 PM, "Tony Yu" <tsyu80(a)gmail.com> wrote: > > While looking through the sphinx config file, I noticed that we have config settings for the `plot_directive` extension and also include the sphinx extension in `doc/ext/`. A quick grep of the doc directory suggests that we don't actually use the plot directive anywhere. Is it worth keeping for future use or should I remove that as part of this doc clean up? There's probably some legacy cruft left over--few people ever read that file, so feel free to clean it up. Is the plot directive non needed to build figures for the user guide? Thanks! Stéfan > > -Tony

1 0

Re: Timing comparisons between mahotas & scikits-image
by Luis Pedro Coelho Sept. 2, 2012

Sept. 2, 2012

Great! I will redo the timings in the next round of manuscript corrections (I already submitted last weeks results yesterday) for the mahotas paper. I might try to speed up things on mahotas' side too :) A little friendly competition always gets users better software. Best, Luis On 09/02/2012 03:41 PM, Stéfan van der Walt wrote: > On Thu, Aug 30, 2012 at 9:46 AM, Luis Pedro Coelho <luis(a)luispedro.org > <mailto:luis@luispedro.org>> wrote: > > I made a timing comparison between mahotas & scikits-image. > > > > operation | mahotas | pymorph | skimage > > ------------+----------+----------+--------- > > dilate | 11.44 | 8.93 | 41.59 > > With the latest versions of skimage and mahotas, I have: > > | mahotas | skimage > ----------+------------+-------- > erode | 9.60 | 16.18 > dilate | 11.69 | 8.01 > open | 20.53 | 16.16 > > Thanks to Johannes Schönberger for the speed-ups. > > Stéfan >

1 0

Re: Numpydoc question
by Tony Yu Sept. 2, 2012

Sept. 2, 2012

On Sun, Sep 2, 2012 at 12:12 PM, Stéfan van der Walt <stefan(a)sun.ac.za>wrote: > On Sun, Sep 2, 2012 at 8:11 AM, Tony Yu <tsyu80(a)gmail.com> wrote: > > I'm inclined to just add `numpydoc_show_class_members = False` to > > `doc/source/conf.py` to reduce the noise during the build. Objections? > > Go ahead--for now, having a clean doc build is more important, until > someone can investigate the issue in more depth. > > Stéfan > While looking through the sphinx config file, I noticed that we have config settings for the `plot_directive` extension and also include the sphinx extension in `doc/ext/`. A quick grep of the doc directory suggests that we don't actually use the plot directive anywhere. Is it worth keeping for future use or should I remove that as part of this doc clean up? -Tony

1 0

Re: Numpydoc question
by Stéfan van der Walt Sept. 2, 2012

Sept. 2, 2012

On Sun, Sep 2, 2012 at 8:11 AM, Tony Yu <tsyu80(a)gmail.com> wrote: > I'm inclined to just add `numpydoc_show_class_members = False` to > `doc/source/conf.py` to reduce the noise during the build. Objections? Go ahead--for now, having a clean doc build is more important, until someone can investigate the issue in more depth. Stéfan

1 0

Numpydoc question
by Tony Yu Sept. 2, 2012

Sept. 2, 2012

Recently, there's been a spike in the number of warnings during the doc build. It turns out that most of these errors are related to class definitions. It just so happens that we've recently had a lot of PRs that added custom classes (Image repr PR, Transforms PR, and Viewer PR). This leads me to my question: Does anyone know the effect of the sphinx config flag: `numpydoc_show_class_members` This flag is on by default, but setting it to False fixes a lot of the build warnings (and also removes many of the ImportErrors you mentioned on the viewer PR, Johannes). The strange thing is that turning this flag off just removes the table of class methods *but not* the normal docs for those methods. (For example, in the MCP class<http://scikits-image.org/docs/dev/api/skimage.graph.html#mcp>, the `Methods` heading and the 2-row table just below it will disappear, but the docs below that table remain.) I'm inclined to just add `numpydoc_show_class_members = False` to `doc/source/conf.py` to reduce the noise during the build. Objections? -Tony

1 0