#25 Variious issues

stable
open
nobody
5
2009-01-07
2009-01-07
No

Configuration: WinXP SP 3, FOP 0.95, JEculid 3.1.4

I runned the entire test suite on my machine here are the results:

1) Vis-à-vis the example provided in the test-suite - and sometimes to common sense - JEuclid gerenates too many or not enough brackets (e.g. 013.pdf)
2) leq rendered with #02266, for consistency with geq use #02264 (013.pdf)
3) conjugate not rendered properly (014.pdf)
4) fn generates overwritten characters (027.pdf)
5) trigonometric functions without variable not rendered (065.pdf)
6) piecewise closing bracket added on the right side (122.pdf)
7) diff not rendered as f' (132.pdf)
8) partialdiff with list not sure about the rencering (149.pdf)
9) outerproduct (264.pdf), scalarproduct (265.pdf) not renderend correctly
10) selector does not render correctly (266.pdf)
11) verticaL arrow direction (288.pdf to 293.pdf), recommendation shows oblique arrow (MathML 4.4.7.3.3)
12) mean is not rendered, MathML 4.4.9.1 (312.pdf)
13) momentabout is not rendered properly (315.pdf), I suspect thereis a typo in the recommendation 4.4.9.7.3 (see also http://mathworld.wolfram.com/Moment.html\) should it not be <X^3>_p
14) notprsubset symbol incorrect (MathML 4.4.6.10.3) (336.pdf)
15) notsubset symbol incorrect (MathML 4.4.6.9.3) (337.pdf) symbol appears construct where unicode characters exist in range #02282-#02289
16) subset symbol incorrect (MathML 4.4.6.7.3) (339.pdf)
17) prsubset symbol incorrect (MathML 4.4.6.7.3) (340.pdf)
18) generated unrecognized token NaN (in ErrorHandling/BadAttribs/badAttribsGlyph4.mml and ErrorHandling/BadChildren/emptyContent1.mml and Topics/WhiteSpace/invChars.mml [692.pdf])
19) mfenced attributes (empty strings or spaces) do not override behaviour (414.pdf)
20) fontweight, fontstyle not processed (font issue?) (433.pdf, 552.pdf)
21) mpadded does not appear to perform as expected (437.pdf - 460.pdf)
22) mmultiscripts does appear to perform as expected (MathML 3.4.7, 484.pdf, 485.pdf)
23) mlabeledtr labels in table (MathML 3.5.3.3, 504.pdf)
24) mtable width parameter is not taken into account (524.pdf, 526.pdf, 527.pdf)
25) mtd rowspan attribute not processed as expected (531.pdf)
26) test Presentation/TokenElements/mglyph/rec-mglyph1.mml causes unrecognized token NaN(538.pdf)
27) mo stretchy attribute ignored (585.pdf)
28) linebreak processing not working (598.pdf, 645.pdf, 647.pdf - 651.pdf)
29) closing bracket ) not stretched in msup (611.pdf)
30) mstyle background attribute ignored (647.pdf, 648.pdf, 649.pdf, 651.pdf)
31) nested script, 1st and 2nd level not very apparent (656.pdf)
32) tilde not stretched (666.pdf)
33) TortureTests/Complexity/complex3.mml, OutOfMemory, dammaged pdf

Comments/suggestions:

i) I prefer the symbol representation of quotient, curl, div, grad
ii) inverse (-1) iso -1 follows the recommendation but is counter-intuitive (http://mathworld.wolfram.com/Inverse.html)
iii) variance, power 2 after bracket iso on sigma (http://mathworld.wolfram.com/Variance.html) (315.pdf)
iv) represent polar expressions as : a e^[imaginary i]b (344.pdf)
v) error processing, no warnings generated in case of errors

File with output pdf files, fop configuration, listfonts output, and detailed log of doubtful cases cases can be provided on request, >1MB, cannot be uploaded in the tracker)

Discussion

  • Max Berger
    Max Berger
    2009-01-07

    Raul,

    the files would be needed, because I have no idea which version of the testsuite you are referring to. Please create a .zip archive and send it to me via email, it should contain:
    - the original files [ or send me a link ]
    - the rendition on your machine (the pdfs you're referring to)
    - if you have, the "should be" output [ or a link ]

    most of your bugs seem to be related to content MathML - this is handled in a XSL stylesheet. If you have XSLT knowledge, please look at [1] at try to provide suggestions / a patch.

    Some of your issues are stylistic issues, where you say you "prefer" a specific representation - this should not be changed generally, but rather through parameters. The stylesheet already contains some parameters for representations, we may add the ones you like (such as quotient, curl, div, grad, ...?).

    Max

    [1] http://jeuclid.svn.sourceforge.net/viewvc/jeuclid/trunk/jeuclid-core/src/main/resources/content/mathmlc2p.xsl?view=markup

     
  • Max,

    I made some changes to the mathmlc2p that should cover the following items: 2, 3, 7(?), 9, 11, 12, 13, 14, 15, 16.
    see attached file mathmlc2p-raul.xsl

    Further observations:

    * item 6: piecewise is rendered using mfenced with attributes, the closed symbol is an empty string. This will be solved by solving item 19.

    * item 7 the stylesheet using the following source code:
    <apply>
    <diff/>
    <ci> g </ci>
    </apply>
    will generate something like:
    <mrow>
    <mi fontweight="bold">
    g
    </mi>
    </mrow>'
    the ' is outside any math tag, would the following output work ????
    <mrow>
    <mrow>
    <mi fontweight="bold">
    g
    </mi>
    </mrow>
    <mo>'</mo>
    </mrow>

     
  • Draft, see comment 2009-01-08 10:42

     
    Attachments
  • Next proposal and examples

     
    Attachments
  • Max,

    Please find a version of the xsl which addresses the symbols related issue (2, 3, 7, 9, 11, 12, 13, 14, 15, 16).
    It also solve the problem of the trigonometric functions which where not rendered when they were not the first child of <apply>.
    Up to now, I only did some simple testing.
    Please note that the conjugate symbol is not stretched.

    Raul,

     
  • Max,

    Please find a new version of mathmlc2p.xsl, this version further corrects the fn display issue.

    Raul

     
  • Version 2009-01-13 09:13

     
    Attachments
  • Any new concerning the linebreak support?