Tralics, a LaTeX to XML translator; Part II


Note 1. Email:

Note 2. If your browser does not show correctly the math, click on the footnote in order to get an image utf8.png

Note 3. The \uppercase command cannot be expanded; however its evaluation is a sequence of tokens that will be read again, expanded, and evaluated.

Note 4. The first character should be X or a digit, otherwise, the XML source is invalid. If you replace xe9 by xy9 or ye9, strange errors may be signaled; the case of a non-ASCII character is worse

Note 5. These definitions of less-than and greater-than are wrong; they will be redefined later.

Note 6. The code on lines 70–79 overrides some of these settings

Note 7. The trouble is that, when the aux file contains \newlabel, its argument uses \csname, this expands to é, and this does not match the UTF-8 character.

Note 8. Guess: why is there no \expandafter after the second \else?

Note 9. Here \do is short for \XML@doattribute

Note 10. Here \Att is short for \XML@attrib

Note 11. This is unclear to me

Note 12. This is redefined by \raggedbottom to be empty; you have to find a different trick if you want to insert something at a given position using this code as a model.

Note 13. and not the subtheme; we could use a command that picks up the first three characters.

Note 14. URL is, see also slices 1 to 7.

Note 15. The English language is selected by: GB, US, gb, us, en, and none. But not EN. New in V2: en_GB and en_US.

Note 16. I don´t understand this \temp stuff. Moreover, as we shall see, we never use the \label command

Note 17. The kludge is still active in 2006; let´s hope it will be removed in 2007.

Note 18. Is it possible to know, in TeX, if a given page is the last in a sequence? Maybe, we could modify the \output routine, so that the \clearpage on line 372 will output some pages, the last one being special. The situation is complicated by the fact that lines 377 and 380 may output empty pages in the sequence. The \clearpage on line 389, together with the \BlankPpage on lines 393 and 396 are also candidates for producing the last page in the sequence.

Note 19. This last one explains how page numbers are typeset; see later.

Note 20. code shown here as `...´, it is a bit more complicated in V2

Note 21. It seems that fotex does not use absolute values.

Note 22. They are unused by fotex.

Note 23. The \vskip commands, commented out in V2 as “not needed” on lines 1253 and 1260, have as side effect to end the paragraph. This gives unexpected side effects. In the Raweb case, cells are not directly in a table.

Note 24. What if the page footer contains a table with a blue cell?. Note that the \clearpage on line 1220 forbids putting tables inside page headers...

Note 25. We hacked a bit this code.

Note 26. Why do we add the indentation?

Note 27. In French, accueil means reception, welcome.

Note 28. Fondements = foundations

Note 29. Logiciel = software

Note 30. `LeProjet´ is French for `TheTeam´.

Note 31. Here the French word développé is used in the sense of expanded. All tag names are ASCII 7bits, although XML allows any Unicode character.

Note 32. The French word `habilité´, with an acute accent, means `entitled´, hence `habilité à diriger des recherches´, in short HDR, means entitled to be a Phd thesis supervisor

Note 33. This should be removed for 2005. The error message changed.

Note 34. There is no apply-templates here; this is wrong (JG).

Note 35. A paragraph can be centered, as well as a cell in table, but not inline elements like <hi>.

Note 36. This is strange; the test should be done on the rows, and all borders.

Note 37. This seems to be a bug. Tralics converts both `table´ and `tabular´ environments to <table>, in the first case, there is no caption.

Note 38. The result is so complicated that the XSL/Format style sheet ignores this attribute. This should be changed in 2005.

Note 39. Why not a simple copy-of ?

Note 40. This is strange; Tralics does not produce such a thing.

Note 41. See for instance

Note 42. See for instance

Note 43. See for instance; this is the version without the TOC.

Note 44. If the acceskey is, say N, pressing down the alt key together with the letter N has the same effect as clicking on the icon, at least with my browser.

Note 45. The `small´ was removed in 2006

Note 46. The order of these two items changed between 2005 and 2006

Note 47. This is new in 2005; but theme_NUM.html and theme_Num.html are handled the same by the Web server.

Note 48. This is a strange name. It has no obvious meaning

Note 49. We should test the value.

Note 50. Anchors were added in version 2.8.5, in order to make \index work.

Note 51. Question: what says the DTD about this style attribute?

Note 52. What if the initial XML document specifies a different horizontal alignment, something hidden in the style attribute?

Note 53. What happens if both style and align are given? who wins? In fact, the question has no sense, because Tralics never emits a `align´, so that this template shoud never be called.

Note 54. Or an anchor, since version 2.8.5

Note 55. This sets the target attribute of the anchor. This behaviour is sometimes annoying.

Note 56. If you look at the code line 487, you can see that conversion from one DTD to the other does not add ids to formulas. Hence, if Tralics adds no useless ids, you will see the same equation number in the Pdf and the HTML.

Note 57. The first section is numbered zero for no obvious reason, so that `fondements´ is numbered 3 in the Pdf filem and 2 in the HTML file.

Note 58. Why is there a space before the number? The LaTeX equivalent would be wrong. It seems that my brower discards these spaces.

Note 59. According to M. Kay, in XSLT2.0, if the style sheet starts with `version=2.0´, all id attributes of all modules are concatenated, and the result is a link to nowhere; if the style sheet has version=1.0, only the id is used. In any case, it would be better to check that there is a single module that matches, and make the link to it.

Note 60. The bibliography part of this DTD comes frome the TEI. See

Note 61. What about part and junior? Note that Tralics generates an empty part attribute, that is omitted from the element, but junior should be added to <surname>.

Note 62. in the sense of `sorted´; secret documents should not appear in the Raweb.

Note 63. The value of the link is in xlink. It happens that Tralics puts the same value in the link and in the attribute in most of the cases. But this is liable to change.

Note 64. It is removed in the 2005 version shown here.

Note 65. This implies that the first character has to be lower-casified

Note 66. What about junior?

Note 67. In the <team> element, the <hdr> element is taken into account, same method as in the HTML case, see previous chapter.

Note 68. This is a bug in Tralics, of course

Note 69. Why not use a random number? there should be a parameter in Tralics that inhibits this template.

Note 70. Normal text font, but XML-like color

Note 71.

Note 72. How is this attribute defined?

Note 73. It is too complicated to explain here

Note 74. This is a strange name. Maybe one day, a better one will be used


Back to main page