Mailman 3 July 2009 - SciPy-Dev

Re: [SciPy-dev] Category of the week
by David Goldsmith 24 Jul '09

24 Jul '09

--- On Thu, 7/23/09, Ralf Gommers <ralf.gommers(a)googlemail.com> wrote: > Fourier Transforms or Other Random Ops? Well, I just polished off three of the remaining eight in ORO, so let's finish that one off, and if we do so early, dive in to FT. DG > > Cheers, > Ralf > > > > > If a cat-leader wants to recruit helpers by nominating > their cat, those are: > > > > "Financial Functions" (Skipper Seabold), > "Random Number Generation" (Alan Jackson), and > "Masked Arrays III" (Scott Sinclair). > > > > Finally, our "category" of the week could be > simply to move all "Goal met (almost)"s to > "**Goal Met!**" status; these are: > > > > "Array Manipulation," "Functional > Operations," "Ndarray II," and > "Ufunc." > > > > DG > > > > PS: What's w/ all the uncategorized white and light > grey's one sees on the "All Docstrings" page? > > > > > > > > _______________________________________________ > > Scipy-dev mailing list > > Scipy-dev(a)scipy.org > > http://mail.scipy.org/mailman/listinfo/scipy-dev > > > >

1 0

correlate_nd.src won't compile
by Pierre GM 23 Jul '09

23 Jul '09

All, I just run into the following problem while compiling scipy (0.8.0svn5874). Any comments welcome. Thx. (mac OS 10.5.7, i686-apple-darwin9-gcc-4.2.1 (GCC) 4.2.1 (Apple Inc. build 5566)) creating build/temp.macosx-10.5-i386-2.5/scipy/signal creating build/temp.macosx-10.5-i386-2.5/build/src.macosx-10.5- i386-2.5/scipy/signal compile options: '-Iscipy/signal -I/Users/pierregm/ Computing/.pythonenvs/trash_test/lib/python2.5/site-packages/numpy/ core/include -I/opt/local/Library/Frameworks/Python.framework/Versions/ 2.5/include/python2.5 -c' gcc-4.0: scipy/signal/firfilter.c gcc-4.0: build/src.macosx-10.5-i386-2.5/scipy/signal/correlate_nd.c scipy/signal/correlate_nd.c.src: In function ‘_imp_correlate_nd_ubyte’: scipy/signal/correlate_nd.c.src:129: warning: implicit declaration of function ‘PyArrayNeighborhoodIter_ResetConstant’ scipy/signal/correlate_nd.c.src:133: warning: implicit declaration of function ‘PyArrayNeighborhoodIter_NextConstant’ scipy/signal/correlate_nd.c.src: In function ‘_imp_correlate_nd_ubyte_2d’: scipy/signal/correlate_nd.c.src:133: warning: implicit declaration of function ‘PyArrayNeighborhoodIter_NextConstant2D’ scipy/signal/correlate_nd.c.src: In function ‘_correlate_nd_imp’: scipy/signal/correlate_nd.c.src:287: error: ‘NPY_NEIGHBORHOOD_ITER_ZERO_PADDING’ undeclared (first use in this function) scipy/signal/correlate_nd.c.src:287: error: (Each undeclared identifier is reported only once scipy/signal/correlate_nd.c.src:287: error: for each function it appears in.) scipy/signal/correlate_nd.c.src:287: error: too many arguments to function ‘*(_scipy_signal_ARRAY_API + 852u)’ scipy/signal/correlate_nd.c.src:301: error: too many arguments to function ‘*(_scipy_signal_ARRAY_API + 852u)’ scipy/signal/correlate_nd.c.src: In function ‘_imp_correlate_nd_ubyte’: scipy/signal/correlate_nd.c.src:129: warning: implicit declaration of function ‘PyArrayNeighborhoodIter_ResetConstant’ scipy/signal/correlate_nd.c.src:133: warning: implicit declaration of function ‘PyArrayNeighborhoodIter_NextConstant’ scipy/signal/correlate_nd.c.src: In function ‘_imp_correlate_nd_ubyte_2d’: scipy/signal/correlate_nd.c.src:133: warning: implicit declaration of function ‘PyArrayNeighborhoodIter_NextConstant2D’ scipy/signal/correlate_nd.c.src: In function ‘_correlate_nd_imp’: scipy/signal/correlate_nd.c.src:287: error: ‘NPY_NEIGHBORHOOD_ITER_ZERO_PADDING’ undeclared (first use in this function) scipy/signal/correlate_nd.c.src:287: error: (Each undeclared identifier is reported only once scipy/signal/correlate_nd.c.src:287: error: for each function it appears in.) scipy/signal/correlate_nd.c.src:287: error: too many arguments to function ‘*(_scipy_signal_ARRAY_API + 852u)’ scipy/signal/correlate_nd.c.src:301: error: too many arguments to function ‘*(_scipy_signal_ARRAY_API + 852u)’

4 6

Re: [SciPy-dev] Special assignment for someone w/ "Unimportant" privileges
by David Goldsmith 23 Jul '09

23 Jul '09

--- On Thu, 7/23/09, Ralf Gommers <ralf.gommers(a)googlemail.com> wrote: > There's a few different types of uncategorized > docstrings: > > - some that were overlooked. i think charrayarray and > matrix can be two new categories for example. > - methods of classes that are in a category (see for > example broadcast), those can be left where they are as > well. they can easily be done together with the class, but > there's no need to stuff them in existing categories. Matrix already is a category, one marked Goal Met, but I see that many of class matrix's methods are white or grey: Joe, if you read this, since you set the goal, does it include these methods that aren't on the Milestones page themselves, but are methods of a class that is on the Milestones page? I know the general feeling is that the Milestones page is already too long (i.e., takes too long to load because so many objects on it means database queries take a long time), but, if these methods are part of the goal, I'd like to see these "methods-in-the-goal-but-not-on-the-Milestones-page" all in one place somewhere, sorted out of the other "un-categories" Ralf discusses below. I know the Q & A page is also getting long, but what about there? > - the rst files that form the User Guide, can be left where > they are for now. > > - things like testing and distutils. it's important > that they get documented at some point, but they are not > very accessible topics and not as urgent as many of the > categories. fine where they are. OK, so two of these "un-categories" are "Unimportant Now, Do Eventually"; I know we're just going to have to redo the color scheme again shortly to accommodate the dual review process, but how difficult would it be to add one color now, say lavender, or bright orange, for this "UNDE" status? > - some unimportant ones. there really are not that many of > them. Still, it'd be nice to weed them out, as I (and I'm sure many other potential writers) don't know which they are - I don't want to be working on 'em, and I don't want volunteers working on 'em, if they're Unimportant. > The one big group of docstrings I'm not sure about is > everything in numpy.generic. They are all undocumented and > the source is not available in the wiki. Anyone know what to > do with those? Well, someone "in the know" could take a look at what I did for the attributes in the .generic namespace and opine on whether that's (in)sufficient... DG > > Cheers, > Ralf > > > > > > DG > > > > > > > > _______________________________________________ > > Scipy-dev mailing list > > Scipy-dev(a)scipy.org > > http://mail.scipy.org/mailman/listinfo/scipy-dev > > > > > -----Inline Attachment Follows----- > > _______________________________________________ > Scipy-dev mailing list > Scipy-dev(a)scipy.org > http://mail.scipy.org/mailman/listinfo/scipy-dev >

1 0

Category of the week
by d_l_goldsmith＠yahoo.com 23 Jul '09

23 Jul '09

Hi, again! So, how'd we do on last week's (second) "category" of the week (about NaN treatment in masked arrays, IIRC)? I focused on getting the "Basic Objects" section done, so I really didn't follow progress on the other. Anyway, time for a new one: the remaining (un-led) categories requiring substantial work are: "Fourier Transforms," "Other Random Operations," "Masked Arrays (I) and IV," "Operations on Masks," "Even More MA Functions I & II," "Numpy Internals," "C-types," "Other Math," & "Other Array Subclasses." If a cat-leader wants to recruit helpers by nominating their cat, those are: "Financial Functions" (Skipper Seabold), "Random Number Generation" (Alan Jackson), and "Masked Arrays III" (Scott Sinclair). Finally, our "category" of the week could be simply to move all "Goal met (almost)"s to "**Goal Met!**" status; these are: "Array Manipulation," "Functional Operations," "Ndarray II," and "Ufunc." DG PS: What's w/ all the uncategorized white and light grey's one sees on the "All Docstrings" page?

2 1

Special assignment for someone w/ "Unimportant" privileges
by David Goldsmith 23 Jul '09

23 Jul '09

OK, I promoted a few white ones on the Milestones page to light grey, counted the remainder (most in one or the other of the MA categories) and then cut, paste, and "counted" all the white ones on the "All docstrings" page: the ratio is about 5:1 (300:60). So I look more closely at some of the white ones on the All docstrings page and see things like: numpy.testing.decorators.deprecated and numpy.ndarray.conjugate Shouldn't the former (and many like it) be classified as "Unimportant"? And why wasn't the latter (and many like it) included in the category of its namespace? So, the "Special Assignment" for someone who has the permissions to do this (I don't) is to go through the white ones (at the very least) on the All docstrings page and reclassify as "Unimportant" those for which such a reclassification is appropriate. Let be any that you're unsure of, but compile a list of 'em, and post it at the Q & A page for others to contribute their opinions (or expert decree). As for the categorizable uncategorized ones, unless someone tells me why they weren't categorized, I'll go ahead and do that. DG

2 1

Bar graph of status in Stats
by David Goldsmith 23 Jul '09

23 Jul '09

Hi, folks. Would anyone object to the reversing the (chronological) order of the bar graph showing overall status on the Stats page of the Wiki - at this point I don't really think we care about last summer's performance as much as where we are currently, so it'd be nice if that were the first thing one sees instead of the last. ;-) Is pydocweb written such that this requires only a single (or two at most) change(s) in the code? DG

1 0

Update and "open issues"
by d_l_goldsmith＠yahoo.com 23 Jul '09

23 Jul '09

OK, since no one's RSVP-ed in the affirmative, and no one's Skyping me, I'll just post an "Update and Open Issues": Update ------ 0) Jack Liddle has fulfilled his promise to draft a Scipy "Milestones" page (i.e., analogous to the Numpy "Milestones" page): http://docs.scipy.org/scipy/Milestones/ Nice job, Jack, thanks!!! 1) I've added "Reviewer Guidelines" for both technical and presentation reviewers to the "Questions and Answers" page on the Numpy Wiki (please discuss these there). 2) "Basic Objects" on the Numpy "Milestones" page is all but Goal Met - only two (not Unimportant) items remain to be brought to "Needs Review" status: ndarray.strides (I'll give it the old "college try," but I'd really appreciate it if someone w/ a "deeper" understanding _and_ a talent for clear explanation could try to tackle this one), and numpy.ufunc (i.e., the class itself) about which there's an Open Issue (see below). Open Issues ----------- 0) Presently, _class_ numpy.ufunc is documented as if it were a function, and not to function spec to boot (albeit for reasons which make some sense once you see how it deviates). IMO, it should be documented as the class that it is, but I'm sympathetic to the reason(s) it's done as it presently is, so I'm open to being convinced. However, it's present deviations are inconsistent w/ the pydocweb application, so if we keep it "as is," we need to decide whether we care, and if so, ticket(s) to make the ap compliant need to filed. The rest are courtesy of Ralf: 1) Ralf would "like to be able to do a few more things on the wiki like make things 'Unimportant,' set things to 'Reviewed, Needs Work,' etc. (That way [he doesn't] have to leave comments in the hope Pauli will spot them eventually. [He's] always hesitant to add things to [Pauli's] mountain of TODOs.)" 2) An update on prospects for future funding. (Joe?) (Perhaps I, DG, should add this to the 'Q&A' so that Joe has a convenient way and place to update us all as things develop...) OK, that's what I have; email me anything you want to add. DG

4 3

Re: [SciPy-dev] Update and "open issues"
by David Goldsmith 22 Jul '09

22 Jul '09

Thanks, David. Have you now opened a ticket? DG --- On Wed, 7/22/09, David Warde-Farley <dwf(a)cs.toronto.edu> wrote: > From: David Warde-Farley <dwf(a)cs.toronto.edu> > Subject: Re: [SciPy-dev] Update and "open issues" > To: d_l_goldsmith(a)yahoo.com, "SciPy Developers List" <scipy-dev(a)scipy.org> > Date: Wednesday, July 22, 2009, 1:54 PM > Hi David, > > I mentioned this on the list a while ago but forgot to open > a ticket: > > http://docs.scipy.org/doc/numpy/reference/generated/numpy.where.html?highli… > > Notice the crossreference link to nonzero() is wrong; it > sends you to the C API when it should send you to the Python > function in the same subpackage. I'm not sure how to debug > this, or if there's a systematic way to find other instances > where this is happening and fix them. > > David > > > On 22-Jul-09, at 3:34 PM, d_l_goldsmith(a)yahoo.com > wrote: > > > > > OK, since no one's RSVP-ed in the affirmative, and no > one's Skyping me, I'll just post an "Update and Open > Issues": > > > > Update > > ------ > > > > 0) Jack Liddle has fulfilled his promise to draft a > Scipy "Milestones" page (i.e., analogous to the Numpy > "Milestones" page): > > > > http://docs.scipy.org/scipy/Milestones/ > > > > Nice job, Jack, thanks!!! > > > > 1) I've added "Reviewer Guidelines" for both technical > and presentation > > reviewers to the "Questions and Answers" page on the > Numpy Wiki (please discuss these there). > > > > 2) "Basic Objects" on the Numpy "Milestones" page is > all but Goal Met - only two (not Unimportant) items remain > to be brought to "Needs Review" status: ndarray.strides > (I'll give it the old "college try," but I'd really > appreciate it if someone w/ a "deeper" understanding _and_ a > talent for clear explanation could try to tackle this one), > and numpy.ufunc (i.e., the class itself) about which there's > an Open Issue (see below). > > > > Open Issues > > ----------- > > > > 0) Presently, _class_ numpy.ufunc is documented as if > it were a function, and not to function spec to boot (albeit > for reasons which make some sense once you see how it > deviates). IMO, it should be documented as the class > that it is, but I'm sympathetic to the reason(s) it's done > as it presently is, so I'm open to being convinced. > However, it's present deviations are inconsistent w/ the > pydocweb application, so if we keep it "as is," we need to > decide whether we care, and if so, ticket(s) to make the ap > compliant need to filed. > > > > The rest are courtesy of Ralf: > > > > 1) Ralf would "like to be able to do a few more things > on the wiki like make things 'Unimportant,' set things to > 'Reviewed, Needs Work,' etc. (That way [he doesn't] have to > leave comments in the hope Pauli will spot them eventually. > [He's] always hesitant to add things to [Pauli's] mountain > of TODOs.)" > > > > 2) An update on prospects for future funding. > (Joe?) (Perhaps I, DG, should add this to the > 'Q&A' so that Joe has a convenient way and place to > update us all as things develop...) > > > > > > OK, that's what I have; email me anything you want to > add. > > > > DG > > > > > > > > _______________________________________________ > > Scipy-dev mailing list > > Scipy-dev(a)scipy.org > > http://mail.scipy.org/mailman/listinfo/scipy-dev > >

1 0

Weekly Summer Marathon Skypecon
by David Goldsmith 22 Jul '09

22 Jul '09

Ready and waiting (or I can simply post an update)... DG

1 0

Skypecon in just under 2 hrs.
by David Goldsmith 22 Jul '09

22 Jul '09

1 0