You are viewing a plain text version of this content. The canonical link for it is here.
Posted to fop-dev@xmlgraphics.apache.org by James Tauber <jt...@jtauber.com> on 1999/11/23 18:24:42 UTC
request for FOP tests
I'd like to start a test directory in the FOP module with a series of
formatting object files. I use a handful of files for testing new features
and regression testing. I'm sure some of you have test files you've
developed.
If people send them to me (privately), I'll put them in xml-fop/test-files
so we can share each others tests and build a test suite.
James
Re: request for FOP tests
Posted by Stephan Albers <St...@af-software.de>.
Arved Sandstrom schrieb:
> > If people send them to me (privately), I'll put them in xml-fop/test-files
> > so we can share each others tests and build a test suite.
> >
> Glad you brought that up, as I was just about to suggest it myself. It's
> not like there are huge amounts of FO files floating around anyway. :-)
>
> You've got the one from me for simple-links, but I'm going to expand that
> and clean it up a bit, and get another out to you in a day or two.
We'll will clean up our examples where FOP currently fails (color
between pages, half cutted line), so we will also have some files for
regression test.
Perhaps the table examples that made problems could also be included?
> Speaking of test files, what about PDF output? In other words, people can
> look at a pair: this FO file, acted on by FOP, produces this PDF...or
> should. :-)
That's what renderx does with their product and it really is nice:
http://www.renderx.com/Demo/index.html
Stephan
Re: request for FOP tests
Posted by Arved Sandstrom <Ar...@chebucto.ns.ca>.
On Wed, 24 Nov 1999, James Tauber wrote:
> > Speaking of test files, what about PDF output? In other words, people can
> > look at a pair: this FO file, acted on by FOP, produces this PDF...or
> > should. :-)
>
> Should we include these as part of the source distribution or merely
> somewhere on the xml.apache.org website? I can see things getting pretty
> big, even including just FO files, let alone the PDFs.
>
Somewhere on the website as opposed to with the source. I wasn't assuming
that *any* of this stuff, including the FO files, would be part of the
source distribution.
Arved
Re: request for FOP tests
Posted by James Tauber <jt...@jtauber.com>.
> Speaking of test files, what about PDF output? In other words, people can
> look at a pair: this FO file, acted on by FOP, produces this PDF...or
> should. :-)
Should we include these as part of the source distribution or merely
somewhere on the xml.apache.org website? I can see things getting pretty
big, even including just FO files, let alone the PDFs.
James
Re: request for FOP tests
Posted by James Tauber <jt...@jtauber.com>.
> Oppsss, I must update my knowledge then. Please undo my previous
> comment. You're totally right, I didn't want to offend anyone.. it was
> just my lack of knowledge on other FO-capable software.
>
> BTW, the renderX software is waporware at this point (will it be
> commercial or free?)
>
> Also, where can I find PassiveTeX? same: is it commercial or free?
see http://www.jtauber.com/fop/links/ for this sort of stuff.
> I'm also trying
>
> XML Spec -(my own stylesheet)-> FO -> PDF
>
> for pdf formattation of all the W3C specifications.
Did you read my post to xsl-list? I have already done this. Let me know if
you want me to send my first version to you.
James
Re: request for FOP tests
Posted by James Tauber <jt...@jtauber.com>.
> > I think so, because I want a replacement for TeX, and without math FOP
> > will just be a toy, for scientific publishing. But it may well be that
> > MathML can be done as a transformation to SVG
>
> This is a great idea!
>
> Do you think it would be possible to design a MathML -> SVG XSLT
> transformation sheet?
No.
MathML -> SVG, which as I said is no different to MathML -> XSL areas, is a
full-blown typesetting/composition/layout task. It is the job of a
formatter. You would not write a transformation sheet to do it any more than
you would write a transformation sheet in place of what FOP does.
James
Re: request for FOP tests
Posted by Stefano Mazzocchi <st...@apache.org>.
Sebastian Rahtz wrote:
> I think so, because I want a replacement for TeX, and without math FOP
> will just be a toy, for scientific publishing. But it may well be that
> MathML can be done as a transformation to SVG
This is a great idea!
Do you think it would be possible to design a MathML -> SVG XSLT
transformation sheet?
In fact, this is what it is: you "style" your math structure using a
graphic format. So you could even have your own math style.
Comments?
--
Stefano Mazzocchi One must still have chaos in oneself to be
able to give birth to a dancing star.
<st...@apache.org> Friedrich Nietzsche
Re: request for FOP tests
Posted by Stefano Mazzocchi <st...@apache.org>.
James Tauber wrote:
>
> > > So bottom line is, MathML->SVG or MathML->XSL Areas is practically the
> same
> > > problem.
> >
> > I humbly disagree.
> >
> > MathML is a structure language and all structure languages do not (well,
> > should not!) have any style information built in.
>
> 1. This is irrelevant to my comment. I said that transforming to SVG is
> practically the same problem (ie requires the same processing, algorithms,
> etc) as transforming to XSL Areas. This is true whether the source
> vocabulary is MathML or FooML.
>
> 2. MathML is *both* a structure language and a presentation language. I was
> thinking of FOP supporting the presentation part of MathML.
I understand, then. Sorry, my MathML knowledge is badly outdated.
> > On the other hand, layout languages do not (again, should not) have any
> > structure information besides that of rendering driving.
>
> I disagree with the should not, but that is a separate issue to the issue of
> MathML.
True, totally irrelevant in this discussion.
> > So, while FOs, SVG, X3D are layout languages (or, in other words,
> > "final" languages in the sense that it's a pain in the ass to perform
> > XSLT transformations from here to another language), MathML, MusicML,
> > DocBook, TEI, etc.. are structure-only languages.
>
> This is not true. MathML has a presentation component. Is is both.
Sorry then. I'll go and increase my knowledge on MathML.
> > A stylesheet is the link between structure and layout since it
> > transforms a structure to a layout following the given transformation
> > contract (process called styling).
>
> This again is not true. If it were, there would be no need for FOP. A
> stylesheet does not give you an exact layout. A stylesheet results in a
> general set of instructions (in the form of formatting objects) which then
> have to be interpreted by a composition/formatting/layout engine. It is true
> that a stylesheet is the link but the result of the transformation itself is
> not layout.
Correct, I should have said "layout rendering instructions" instead of
"layout".
> > So, at least in theory, we should go from MathML (MusicML, whateverML)
> > to SVG and have FOP care about "final" languages only.
>
> XSL FOs are not a final language but they are exactly what FOP cares about.
> SVG isn't a final language because it allows for CSS and XSL styling to be
> applied.
> MathML is really two languages, a computational one and a presentational
> one.
>
> At one level, PDF isn't even final because it still need to be interpreted
> to get dots on screen or dots on paper.
>
> The best way to think of it is not structured versus presentation/final but
> as a continuum.
>
> A one end, you have raster formats
> Further along you have basic vector formats
> Further along you have more advanced vector formats like SVG that can still
> have styling applied
> Further along you have specialised presentation formats like MathML or Music
> Notation
> Around this area you also have abstract formatting formats like XSL FOs
> Further along you have pseudo presentation formats like HTML
> Further along you have the more structured formats like DocBook, TEI, the
> other part of MathML, abstract music representations, etc
>
> I hope this clarifies things.
It does.
My analysis was limited, I totally agree. The world is much more complex
than simple (structure/layout) separation, but I think it's a good
design pattern to have a clean separation between structure and layout.
Of course, layout means a language that drives the rendering engine, it
could either be XML based (fo, svg, mathml) or not (pdf, ps, eps).
What I like in the use of XML for publishing is that the design pattern
of context separation is much more evident than in other systems, even
if this is not, of course, driven by the XML syntax itself by the family
of languages/technologies around XML.
Anyway, I apologize for my comment and I totally got James' point: if
MathML contains presentation information, yes, there is no algorithmical
difference between FO, SVG and MathML.
Stefano, which should code more and talk less...
Re: request for FOP tests
Posted by James Tauber <jt...@jtauber.com>.
> > So bottom line is, MathML->SVG or MathML->XSL Areas is practically the
same
> > problem.
>
> I humbly disagree.
>
> MathML is a structure language and all structure languages do not (well,
> should not!) have any style information built in.
1. This is irrelevant to my comment. I said that transforming to SVG is
practically the same problem (ie requires the same processing, algorithms,
etc) as transforming to XSL Areas. This is true whether the source
vocabulary is MathML or FooML.
2. MathML is *both* a structure language and a presentation language. I was
thinking of FOP supporting the presentation part of MathML.
> On the other hand, layout languages do not (again, should not) have any
> structure information besides that of rendering driving.
I disagree with the should not, but that is a separate issue to the issue of
MathML.
> So, while FOs, SVG, X3D are layout languages (or, in other words,
> "final" languages in the sense that it's a pain in the ass to perform
> XSLT transformations from here to another language), MathML, MusicML,
> DocBook, TEI, etc.. are structure-only languages.
This is not true. MathML has a presentation component. Is is both.
> A stylesheet is the link between structure and layout since it
> transforms a structure to a layout following the given transformation
> contract (process called styling).
This again is not true. If it were, there would be no need for FOP. A
stylesheet does not give you an exact layout. A stylesheet results in a
general set of instructions (in the form of formatting objects) which then
have to be interpreted by a composition/formatting/layout engine. It is true
that a stylesheet is the link but the result of the transformation itself is
not layout.
> So, at least in theory, we should go from MathML (MusicML, whateverML)
> to SVG and have FOP care about "final" languages only.
XSL FOs are not a final language but they are exactly what FOP cares about.
SVG isn't a final language because it allows for CSS and XSL styling to be
applied.
MathML is really two languages, a computational one and a presentational
one.
At one level, PDF isn't even final because it still need to be interpreted
to get dots on screen or dots on paper.
The best way to think of it is not structured versus presentation/final but
as a continuum.
A one end, you have raster formats
Further along you have basic vector formats
Further along you have more advanced vector formats like SVG that can still
have styling applied
Further along you have specialised presentation formats like MathML or Music
Notation
Around this area you also have abstract formatting formats like XSL FOs
Further along you have pseudo presentation formats like HTML
Further along you have the more structured formats like DocBook, TEI, the
other part of MathML, abstract music representations, etc
I hope this clarifies things.
James
Re: request for FOP tests
Posted by Stefano Mazzocchi <st...@apache.org>.
James Tauber wrote:
>
> > I think so, because I want a replacement for TeX, and without math FOP
> > will just be a toy, for scientific publishing. But it may well be that
> > MathML can be done as a transformation to SVG
>
> Well that's kind of what formatting in FOP is like anyway. SVG is similar to
> the abstract areas that the format stage of FOP produces. In fact, it would
> be trivial to add a Renderer that produces an SVG file for each page.
>
> So bottom line is, MathML->SVG or MathML->XSL Areas is practically the same
> problem.
I humbly disagree.
MathML is a structure language and all structure languages do not (well,
should not!) have any style information built in.
On the other hand, layout languages do not (again, should not) have any
structure information besides that of rendering driving.
So, while FOs, SVG, X3D are layout languages (or, in other words,
"final" languages in the sense that it's a pain in the ass to perform
XSLT transformations from here to another language), MathML, MusicML,
DocBook, TEI, etc.. are structure-only languages.
A stylesheet is the link between structure and layout since it
transforms a structure to a layout following the given transformation
contract (process called styling).
Other languages do not fall in this category: in HTML, WML, VoiceML the
distinction between structure and layout is not that clear. For example,
a subset of HTML can be used as a structure definition and CSS used as
style contract.
Anyway, rendering SVG makes sense since the renderer doesn't have to
apply any styling contracts. Rendering MathML cannot be done unless an
hardcoded style contract is used to come up with a graphical layout
definition. Which is what SVG is for.
So, at least in theory, we should go from MathML (MusicML, whateverML)
to SVG and have FOP care about "final" languages only.
--
Stefano Mazzocchi One must still have chaos in oneself to be
able to give birth to a dancing star.
<st...@apache.org> Friedrich Nietzsche
Re: request for FOP tests
Posted by James Tauber <jt...@jtauber.com>.
> I think so, because I want a replacement for TeX, and without math FOP
> will just be a toy, for scientific publishing. But it may well be that
> MathML can be done as a transformation to SVG
Well that's kind of what formatting in FOP is like anyway. SVG is similar to
the abstract areas that the format stage of FOP produces. In fact, it would
be trivial to add a Renderer that produces an SVG file for each page.
So bottom line is, MathML->SVG or MathML->XSL Areas is practically the same
problem.
James
Re: request for FOP tests
Posted by Sebastian Rahtz <se...@computing-services.oxford.ac.uk>.
Stefano Mazzocchi writes:
> I admit my total ignorance on TeX and related stuff (I come from the
> windows dark side of things) and I don't know how it works and how it is
> modularized.
modularized is something it is NOT...
> But I saw a very nice FOP should handle too: MathML! Even if I bet it's
> going to be damn hard.
its easy for TeX, of course, as TeX has a very sophisticated math
formatter built in already. it only took me a day or two to add MathML
to PassiveTeX, and it more or less works with presentation MathML
> Well, maybe the W3C guys that did Amaya (I met one of them a couple of
> years ago) could be interested in joining. Do you want me to ask around?
> Do you think such thing would be a valuable addiction to FOP at all?
I think so, because I want a replacement for TeX, and without math FOP
will just be a toy, for scientific publishing. But it may well be that
MathML can be done as a transformation to SVG
Sebastian
Re: request for FOP tests
Posted by James Tauber <jt...@jtauber.com>.
> But I saw a very nice FOP should handle too: MathML! Even if I bet it's
> going to be damn hard.
>
> Well, maybe the W3C guys that did Amaya (I met one of them a couple of
> years ago) could be interested in joining. Do you want me to ask around?
> Do you think such thing would be a valuable addiction to FOP at all?
Absolutely. SVG and MathML were the two "extensions" I always envisaged FOP
having. I'll ask the MathML people if anyone is interested.
James Tauber
Re: request for FOP tests
Posted by Stefano Mazzocchi <st...@apache.org>.
Sebastian Rahtz wrote:
>
> Stefano Mazzocchi writes:
> >
> > BTW, the renderX software is waporware at this point (will it be
> > commercial or free?)
>
> its not vaporware. it exists. it is neither commercial nor free (in
> the senses I think you mean) -- they are only interesting in providing
> it as part of larger solutions.
I understand.
> > Also, where can I find PassiveTeX? same: is it commercial or free?
>
> http://users.ox.ac.uk/~rahtz/passivetex. Don't get excited! Its
> entirely free, but you need TeX installed
I admit my total ignorance on TeX and related stuff (I come from the
windows dark side of things) and I don't know how it works and how it is
modularized.
But I saw a very nice FOP should handle too: MathML! Even if I bet it's
going to be damn hard.
Well, maybe the W3C guys that did Amaya (I met one of them a couple of
years ago) could be interested in joining. Do you want me to ask around?
Do you think such thing would be a valuable addiction to FOP at all?
--
Stefano Mazzocchi One must still have chaos in oneself to be
able to give birth to a dancing star.
<st...@apache.org> Friedrich Nietzsche
Re: request for FOP tests
Posted by Sebastian Rahtz <se...@computing-services.oxford.ac.uk>.
Stefano Mazzocchi writes:
>
> BTW, the renderX software is waporware at this point (will it be
> commercial or free?)
its not vaporware. it exists. it is neither commercial nor free (in
the senses I think you mean) -- they are only interesting in providing
it as part of larger solutions.
> Also, where can I find PassiveTeX? same: is it commercial or free?
http://users.ox.ac.uk/~rahtz/passivetex. Don't get excited! Its
entirely free, but you need TeX installed
>
> I welcome your different perspective and I think they are not totally
> orthogonal, don't you think?
>
no, I am sure we don't disagree.
sebastian
Re: request for FOP tests
Posted by Stefano Mazzocchi <st...@apache.org>.
Sebastian Rahtz wrote:
>
> Stefano Mazzocchi writes:
>
> > It would be good if you had a reference implementation. But FOP is
> > considered to be the only one around.
>
> This is hard on the boys at RenderX, who have done some brilliant
> work, and on PassiveTeX, which probably does a wider range of FOs than
> FOP today. Although I would not lay any claim to correctness, of course....
Oppsss, I must update my knowledge then. Please undo my previous
comment. You're totally right, I didn't want to offend anyone.. it was
just my lack of knowledge on other FO-capable software.
BTW, the renderX software is waporware at this point (will it be
commercial or free?)
Also, where can I find PassiveTeX? same: is it commercial or free?
I apologize again for my ignorance. :(
> > Maybe something in between, like, for example, create an XSLT/FO
> > stylesheet for the Webster XML dictionary (which is 58Mb of XML!!) and
> > see how FOP deals with that :-)
>
> what are you trying to test? size of file? complexity of FO? maximum
> size of page sequence? Its likely that Webster would need a fairly
> simple .xsl stylesheet, and would just need a LARGE chunk of memory.
I'm also trying
XML Spec -(my own stylesheet)-> FO -> PDF
for pdf formattation of all the W3C specifications. This requires a lot
of XSLT processing but not much FO power, I admit. These are just text
stuff. But I'm targetting on this first.
Of course, latest goal is to provide some high steam DTP power using FO
+ SVG + bitmap, but this is not my current target.
> Personally, I am 100% interested in FOP's coverage of more of the XSL
> spec, and have minimal interest in long documents and/or processing
> speed. If the formatting algorithms are right, performance will follow
> on.
I welcome your different perspective and I think they are not totally
orthogonal, don't you think?
--
Stefano Mazzocchi One must still have chaos in oneself to be
able to give birth to a dancing star.
<st...@apache.org> Friedrich Nietzsche
Re: request for FOP tests
Posted by Sebastian Rahtz <se...@computing-services.oxford.ac.uk>.
Stefano Mazzocchi writes:
> It would be good if you had a reference implementation. But FOP is
> considered to be the only one around.
This is hard on the boys at RenderX, who have done some brilliant
work, and on PassiveTeX, which probably does a wider range of FOs than
FOP today. Although I would not lay any claim to correctness, of course....
> Maybe something in between, like, for example, create an XSLT/FO
> stylesheet for the Webster XML dictionary (which is 58Mb of XML!!) and
> see how FOP deals with that :-)
what are you trying to test? size of file? complexity of FO? maximum
size of page sequence? Its likely that Webster would need a fairly
simple .xsl stylesheet, and would just need a LARGE chunk of memory.
Personally, I am 100% interested in FOP's coverage of more of the XSL
spec, and have minimal interest in long documents and/or processing
speed. If the formatting algorithms are right, performance will follow
on.
Sebastian
Re: request for FOP tests
Posted by Eric SCHAEFFER <es...@posterconseil.com>.
They are developing it, but it isn't finished (I wanted to buy it...).
Their examples are very nice, effectively.
Eric
_______________________________________
Eric SCHAEFFER
eschaeffer@posterconseil.com
POSTER CONSEIL
118 rue de Tocqueville
75017 PARIS
FRANCE
Tel. : 33-140541058
Fax : 33-140541059
_______________________________________
----- Original Message -----
From: Stephan Albers <St...@af-software.de>
To: <fo...@xml.apache.org>
Sent: Wednesday, November 24, 1999 8:12 PM
Subject: Re: request for FOP tests
> > It would be good if you had a reference implementation. But FOP is
> > considered to be the only one around.
>
> There is one commercial implementation, that has some nice examples of
> FO and PDF output:
>
> http://www.renderx.com/Demo/index.html
>
> Stephan
>
Re: request for FOP tests
Posted by Stephan Albers <St...@af-software.de>.
> It would be good if you had a reference implementation. But FOP is
> considered to be the only one around.
There is one commercial implementation, that has some nice examples of
FO and PDF output:
http://www.renderx.com/Demo/index.html
Stephan
Re: request for FOP tests
Posted by Stefano Mazzocchi <st...@apache.org>.
Arved Sandstrom wrote:
>
> On Tue, 23 Nov 1999, James Tauber wrote:
>
> > I'd like to start a test directory in the FOP module with a series of
> > formatting object files. I use a handful of files for testing new features
> > and regression testing. I'm sure some of you have test files you've
> > developed.
> >
> > If people send them to me (privately), I'll put them in xml-fop/test-files
> > so we can share each others tests and build a test suite.
> >
> Glad you brought that up, as I was just about to suggest it myself. It's
> not like there are huge amounts of FO files floating around anyway. :-)
No shit. I couldn't any really good one to distribute with Cocoon and
wrote my simple little test one which sucks.
> You've got the one from me for simple-links, but I'm going to expand that
> and clean it up a bit, and get another out to you in a day or two.
>
> Speaking of test files, what about PDF output? In other words, people can
> look at a pair: this FO file, acted on by FOP, produces this PDF...or
> should. :-)
It would be good if you had a reference implementation. But FOP is
considered to be the only one around.
> And finally, maybe a directory for example PDF's? Somebody posts a
> request, saying, I'd like to see a PDF that does this, or contains that,
> and someone with the appropriate tools creates it and posts it. After all,
> probably not every contributor here is sporting Exchange and pdfTeX and
> all that other good stuff.
This would speed up FO learning by a big amount. I agree. But I wouldn't
want to see this list acting as a FO producer :)
Maybe something in between, like, for example, create an XSLT/FO
stylesheet for the Webster XML dictionary (which is 58Mb of XML!!) and
see how FOP deals with that :-)
--
Stefano Mazzocchi One must still have chaos in oneself to be
able to give birth to a dancing star.
<st...@apache.org> Friedrich Nietzsche
Re: request for FOP tests
Posted by Arved Sandstrom <Ar...@chebucto.ns.ca>.
On Tue, 23 Nov 1999, James Tauber wrote:
> I'd like to start a test directory in the FOP module with a series of
> formatting object files. I use a handful of files for testing new features
> and regression testing. I'm sure some of you have test files you've
> developed.
>
> If people send them to me (privately), I'll put them in xml-fop/test-files
> so we can share each others tests and build a test suite.
>
Glad you brought that up, as I was just about to suggest it myself. It's
not like there are huge amounts of FO files floating around anyway. :-)
You've got the one from me for simple-links, but I'm going to expand that
and clean it up a bit, and get another out to you in a day or two.
Speaking of test files, what about PDF output? In other words, people can
look at a pair: this FO file, acted on by FOP, produces this PDF...or
should. :-)
And finally, maybe a directory for example PDF's? Somebody posts a
request, saying, I'd like to see a PDF that does this, or contains that,
and someone with the appropriate tools creates it and posts it. After all,
probably not every contributor here is sporting Exchange and pdfTeX and
all that other good stuff.
Just some thoughts.
Arved