\everyaftershipout isn't "called" for every pagebreak
Hi, I'm using \appendtoks...\to\everyaftershipout to reset a variable on every pagebreak. This doesn't work as I expect(ed) it, since it seems to only trigger after a paragraph. The following example shows the problem. On every new page, the first footnote should be "New", which is not the case here. What would be the way to go for me, to get the result I described? Thanks in advance, Andreas. The example: \define\test{\doifelse{\getvalue{testvar}}{page}{% \footnote{Repeat}% }{% \footnote{New}% \setvalue{testvar}{page}% }} \appendtoks \global\setvalue{testvar}{} \to\everyaftershipout \starttext \dorecurse{10}{\dorecurse{250}{text }\test} \dorecurse{10}{\dorecurse{250}{text }\test} \stoptext
Am 16.01.2010 um 17:53 schrieb Andreas Schneider:
Hi,
I'm using \appendtoks...\to\everyaftershipout to reset a variable on every pagebreak. This doesn't work as I expect(ed) it, since it seems to only trigger after a paragraph. The following example shows the problem. On every new page, the first footnote should be "New", which is not the case here. What would be the way to go for me, to get the result I described?
This can't work because TeX collects mote material than it needs for the current page and the same happens with the footnotes, the only thing you can do is to use two pass data Wolfgang
Wolfgang Schuster wrote:
This can't work because TeX collects mote material than it needs for the current page and the same happens with the footnotes, the only thing you can do is to use two pass data
Wolfgang
Is there not even some event in Lua that could be used to handle this case? I think MkIV compatibility is all I need. In case this is also not possible: can you give me a hint where to learn more about twopass processing? Is there in MkIV still something that requires two (or more) passes that I could take a look at to learn how to do it? (I guess I can understand Lua easier than plain TeX, so I guess MkII is out of the picture.) Thanks for your fast answer! Best regards, Andreas.
On 16-1-2010 22:58, Andreas Schneider wrote:
Wolfgang Schuster wrote:
This can't work because TeX collects mote material than it needs for the current page and the same happens with the footnotes, the only thing you can do is to use two pass data
Wolfgang
Is there not even some event in Lua that could be used to handle this case? I think MkIV compatibility is all I need. In case this is also not possible: can you give me a hint where to learn more about twopass processing? Is there in MkIV still something that requires two (or more) passes that I could take a look at to learn how to do it? (I guess I can understand Lua easier than plain TeX, so I guess MkII is out of the picture.)
It depends on your document. We have workflows where we can do with one pass in which case we run with --once (--runs=2 or so) Future versions of context mkiv might need one run less (in some cases) or one perceived run (which saves startup cycles) but as Wolgangs explained already, in more complex cases we cannot do with less runs as any change haping now can influence something a few pages back which in turn might influence what follows. Normally, when you correct only typos there will be no multiple runs. MkIV already is more efficient as we now always assume an advanced pdf backend so that we don't need to use the MkII the object reference mechanism. Hans ----------------------------------------------------------------- Hans Hagen | PRAGMA ADE Ridderstraat 27 | 8061 GH Hasselt | The Netherlands tel: 038 477 53 69 | fax: 038 477 53 74 | www.pragma-ade.com | www.pragma-pod.nl -----------------------------------------------------------------
Hans Hagen wrote:
It depends on your document. We have workflows where we can do with one pass in which case we run with --once (--runs=2 or so)
Future versions of context mkiv might need one run less (in some cases) or one perceived run (which saves startup cycles) but as Wolgangs explained already, in more complex cases we cannot do with less runs as any change haping now can influence something a few pages back which in turn might influence what follows.
Normally, when you correct only typos there will be no multiple runs.
MkIV already is more efficient as we now always assume an advanced pdf backend so that we don't need to use the MkII the object reference mechanism.
Hans
In my example I can't even get \pagenumber to work correctly. If I replace my \test macro with \pagenumber, the first 4 pages all print page "1" (since it's still the same paragraph). Is there anything I can do about that? (Of course my main concern is the original problem of not catching page breaks to reset a variable). I also can't quite work out a multi-pass solution ... the only way I could imagine how it could work is by storing the content I would like to print together with the coordinates it was printed to in the first run, and in the second run evaluate from these information when a page break occured and modify the data accordingly. But I'm a bit lost here where to start ... my TeX knowledge is rather basic and I don't know anything about the luatex internals so I don't know if there are any events I could hook (and when and where to hook them :-/). I would be very grateful if you could point me in a direction I can work forward from - I can only guess that there are similar mechanisms in the already existing MkIV code ... although the \pagenumber problem seems to indicate that it might be near to impossible :-/ Best Regards, Andreas.
On 17-1-2010 22:16, Andreas Schneider wrote:
I would be very grateful if you could point me in a direction I can work forward from - I can only guess that there are similar mechanisms in the already existing MkIV code ... although the \pagenumber problem seems to indicate that it might be near to impossible :-/
dealing with multipass data in mkiv is completely different from mkii and at some point there will be a proper api for users (makes no sense now as i might as well change internals along with luatex developments) in mkiv we store info in tables and much is available during a document run anyway (in mkii we filter data from the tuo file) so the question is, what info do you need? it's probably easier to provide it as core functionality so i need speccs and a small test file Hans ----------------------------------------------------------------- Hans Hagen | PRAGMA ADE Ridderstraat 27 | 8061 GH Hasselt | The Netherlands tel: 038 477 53 69 | fax: 038 477 53 74 | www.pragma-ade.com | www.pragma-pod.nl -----------------------------------------------------------------
dealing with multipass data in mkiv is completely different from mkii and at some point there will be a proper api for users (makes no sense now as i might as well change internals along with luatex developments)
in mkiv we store info in tables and much is available during a document run anyway (in mkii we filter data from the tuo file) so the question is, what info do you need? it's probably easier to provide it as core functionality
so i need speccs and a small test file
Hans
Thanks for that info! I think for specs my "plan" is too vague. I will explain what my goal is then you either see a good solution and/or tell me that this is out of scope for now :-) The test file to demonstrate the problem is in my initial newsgroup post for this thread. I will re-attach it here. What I need would be a precise page break handling. Currently ConTeXt seems to process a whole paragraph as one object, and then breaks it down to individual pages. That leads to a problem however: I want to refine my citation style. All bibliography is inserted into footnotes. If two (or more) consecutive citations _on the same page_ refer to the same bib entry, the footnote for that citation should read "Ibidem". Therefore I have to remember the last cited entry. Currently I use \appendtoks...\to\everyaftershipout to reset/clear the variable I use for storing that last entry. That works fine as long as the paragraphs are small. If however several citations within one (possibly large) paragraph are separated by a page break, the output will be wrong, since the event to reset the variable occurs too late (see above). The problem can be demonstrated with this example - the first footnote on each page should be "New". \define\test{\doifelse{\getvalue{testvar}}{page}{% \footnote{Repeat}% }{% \footnote{New}% \setvalue{testvar}{page}% }} \appendtoks \global\setvalue{testvar}{} \to\everyaftershipout \starttext \dorecurse{10}{\dorecurse{250}{text }\test} \dorecurse{10}{\dorecurse{250}{text }\test} \stoptext Best Regards, Andreas.
2010/1/17 Andreas Schneider
What I need would be a precise page break handling. Currently ConTeXt seems to process a whole paragraph as one object, and then breaks it down to individual pages. That leads to a problem however:
This is how TeX works. :-) Best Martin
On 17-1-2010 23:13, Andreas Schneider wrote:
dealing with multipass data in mkiv is completely different from mkii and at some point there will be a proper api for users (makes no sense now as i might as well change internals along with luatex developments)
in mkiv we store info in tables and much is available during a document run anyway (in mkii we filter data from the tuo file) so the question is, what info do you need? it's probably easier to provide it as core functionality
so i need speccs and a small test file
Hans
Thanks for that info! I think for specs my "plan" is too vague. I will explain what my goal is then you either see a good solution and/or tell me that this is out of scope for now :-)
The test file to demonstrate the problem is in my initial newsgroup post for this thread. I will re-attach it here.
What I need would be a precise page break handling. Currently ConTeXt seems to process a whole paragraph as one object, and then breaks it down to individual pages. That leads to a problem however: I want to refine my citation style. All bibliography is inserted into footnotes. If two (or more) consecutive citations _on the same page_ refer to the same bib entry, the footnote for that citation should read "Ibidem". Therefore I have to remember the last cited entry. Currently I use \appendtoks...\to\everyaftershipout to reset/clear the variable I use for storing that last entry. That works fine as long as the paragraphs are small. If however several citations within one (possibly large) paragraph are separated by a page break, the output will be wrong, since the event to reset the variable occurs too late (see above).
The problem can be demonstrated with this example - the first footnote on each page should be "New".
\define\test{\doifelse{\getvalue{testvar}}{page}{% \footnote{Repeat}% }{% \footnote{New}% \setvalue{testvar}{page}% }}
\appendtoks \global\setvalue{testvar}{} \to\everyaftershipout
\starttext \dorecurse{10}{\dorecurse{250}{text }\test}
\dorecurse{10}{\dorecurse{250}{text }\test} \stoptext
as it takes less time wrting a helper than explaining how to do it i just added it as feature (a few years from probably only wolfgang remembers that it's there) i'll upload a beta to the website that provides ... \starttext \dorecurse {100} { test \footnote{\doifnoteonsamepageelse[footnote]{ibidem}{aaa}} } \stoptext keep in mind that such features, depending on multiple passes, might result in a few more as it will never get things right the first time actually, in some cases it might make sense to delete the tuc file before a run Hans ----------------------------------------------------------------- Hans Hagen | PRAGMA ADE Ridderstraat 27 | 8061 GH Hasselt | The Netherlands tel: 038 477 53 69 | fax: 038 477 53 74 | www.pragma-ade.com | www.pragma-pod.nl -----------------------------------------------------------------
Hans Hagen wrote:
as it takes less time wrting a helper than explaining how to do it i just added it as feature (a few years from probably only wolfgang remembers that it's there)
i'll upload a beta to the website that provides ...
\starttext
\dorecurse {100} { test \footnote{\doifnoteonsamepageelse[footnote]{ibidem}{aaa}} }
\stoptext
keep in mind that such features, depending on multiple passes, might result in a few more as it will never get things right the first time
actually, in some cases it might make sense to delete the tuc file before a run
Hans
Thank you very much for all that effort! It works very well and as a side- effect also gives me a nice point to start learning more about the (lua)tex internals. Seems like I will have to go to one of the next ConTeXt user meetings so I can repay you with the beer(s) I now owe you :-) Just in case someone else needs it: I used the following code now to reset my variable whenever a footnote is placed on a new page. That was necessary so a manually set footnote doesn't interfere with my citation-footnotes. (Otherwise, if a manually inserted footnote would be the first on a page, the next citation footnote would evaluate as "being on the same page as the last one".) \appendtoks \doifnoteonsamepageelse[footnote]{}{\global\setvalue{mycitelast}{}} \to\everyinsidenoteinsert Best Regards (and many thanks again!) Andreas.
participants (4)
-
Andreas Schneider -
Hans Hagen -
Martin Schröder -
Wolfgang Schuster