xgridfit-users

[Xgridfit-users] Xgridfit 2.2

[Peter B. <b....@gm...>]

In Xgridfit 2.2 I have started to migrate the executable files (so far
xgridfit, xgfconfig, ttx2xgf) from Bash to Python scripting. The
command-line syntax for xgridfit and ttx2xgf is as before, but long
option names are available in addition to the short ones.

Instead of knowing about just one XSLT processor and one validator,
which it always looks for, xgridfit maintains a list of eight, and
executes the first one it finds. Xgfconfig reorders the lists of
processors and validators: see the man page for details. You can
temporariy override your standard choice of processor and validator
using command line options.

Xgridfit now validates against a RELAX NG schema by default; use the
--skip-validation command line option to skip.

Several bugs have been fixed (thanks, Andrey!).

There used to be significant limitations on how macros and called
glyph programs could call each other. These limitations have largely
been lifted: macros can even recurse. This revision is quite different
from the fix I posted a few days ago, introducing a lot more
flexibility.

Peter

http://xgridfit.sf.net/

[Xgridfit-users] Xgridfit 2.2

[Peter B. <b....@gm...>]

I forgot to mention that Xgridfit 2.2 requires the Python bindings for
libxml2. I believe that all the usual Linux repositories have it.

PB

Re: [Xgridfit-users] Xgridfit 2.2

[Andrey V. P. <pa...@ca...>]

I have several issues with Xgridfit 2.2:

1. A "--root $(DESTDIR)/" argument should be added to "python setup.py install" 
command in Makefile.

2. The build of Istok font http://code.google.com/p/istok/ is broken with error 
messages:

line 0: Extra element pre-program in interleave
line 1039: Element xgridfit failed to validate content
Error validating document

3. The build of Heuristica font http://code.google.com/p/evristika/

xgridfit -m -p 25 -G no -i Heuristica-Regular_.sfd -o Heuristica-Regular.ttf -O 
Heuristica-Regular.py Heuristica-Regular.ed.xgf

is broken with error message:

execv() arg 2 must contain only strings

4. "-S ..." option to xgridfit does not work as expected even with xsltproc.

-- 
Andrey V. Panov
http://canopus.iacp.dvo.ru/~panov/

Re: [Xgridfit-users] Xgridfit 2.2

[Peter B. <b....@gm...>]

Xgridfit now validates all program files by default. To skip this
step, include an option --skip-validation. However, I don't recommend
that, since invalid program files may cause problems.

I'm having trouble running the Makefile for Istok:

ttx2xgf Istok-Regular.gen.ttx Istok-Regular.gen.xgf
xmllint --xinclude Istok-Regular.ed.xgf > Istok-Regular.tmp.xgf
warning: failed to load external entity "Istok-Regular_acc.xgf"
Istok-Regular.ed.xgf:748: element include: XInclude error : could not
load Istok-Regular_acc.xgf, and no fallback was found
make: *** [Istok-Regular.tmp.xgf] Error 1
rm Istok-Regular.tmp.xgf Istok-Regular.gen.ttx Istok-Regular.gen.xgf

Is this a different problem?

The problem with Heuristika looks like a Python problem: I'll figure
it out, but for now it will work with --processor=libxslt.

I'll work on the -S and --root things today.


On Tue, Mar 30, 2010 at 12:00 AM, Andrey V. Panov
<pa...@ca...> wrote:
> I have several issues with Xgridfit 2.2:
>
> 1. A "--root $(DESTDIR)/" argument should be added to "python setup.py install"
> command in Makefile.
>
> 2. The build of Istok font http://code.google.com/p/istok/ is broken with error
> messages:
>
> line 0: Extra element pre-program in interleave
> line 1039: Element xgridfit failed to validate content
> Error validating document
>
> 3. The build of Heuristica font http://code.google.com/p/evristika/
>
> xgridfit -m -p 25 -G no -i Heuristica-Regular_.sfd -o Heuristica-Regular.ttf -O
> Heuristica-Regular.py Heuristica-Regular.ed.xgf
>
> is broken with error message:
>
> execv() arg 2 must contain only strings
>
> 4. "-S ..." option to xgridfit does not work as expected even with xsltproc.
>
> --
> Andrey V. Panov
> http://canopus.iacp.dvo.ru/~panov/
>
> ------------------------------------------------------------------------------
> Download Intel&#174; Parallel Studio Eval
> Try the new software tools for yourself. Speed compiling, find bugs
> proactively, and fine-tune applications for parallel performance.
> See why Intel Parallel Studio got high marks during beta.
> http://p.sf.net/sfu/intel-sw-dev
> _______________________________________________
> Xgridfit-users mailing list
> Xgr...@li...
> https://lists.sourceforge.net/lists/listinfo/xgridfit-users
>

Re: [Xgridfit-users] Xgridfit 2.2

[Peter B. <b....@gm...>]

This patch will fix problem #3--as I said, a problem with the Python
code. Or grab the updated file from the CVS repository.

On Tue, Mar 30, 2010 at 12:00 AM, Andrey V. Panov
<pa...@ca...> wrote:
> I have several issues with Xgridfit 2.2:
>
> 1. A "--root $(DESTDIR)/" argument should be added to "python setup.py install"
> command in Makefile.
>
> 2. The build of Istok font http://code.google.com/p/istok/ is broken with error
> messages:
>
> line 0: Extra element pre-program in interleave
> line 1039: Element xgridfit failed to validate content
> Error validating document
>
> 3. The build of Heuristica font http://code.google.com/p/evristika/
>
> xgridfit -m -p 25 -G no -i Heuristica-Regular_.sfd -o Heuristica-Regular.ttf -O
> Heuristica-Regular.py Heuristica-Regular.ed.xgf
>
> is broken with error message:
>
> execv() arg 2 must contain only strings
>
> 4. "-S ..." option to xgridfit does not work as expected even with xsltproc.
>
> --
> Andrey V. Panov
> http://canopus.iacp.dvo.ru/~panov/
>
> ------------------------------------------------------------------------------
> Download Intel&#174; Parallel Studio Eval
> Try the new software tools for yourself. Speed compiling, find bugs
> proactively, and fine-tune applications for parallel performance.
> See why Intel Parallel Studio got high marks during beta.
> http://p.sf.net/sfu/intel-sw-dev
> _______________________________________________
> Xgridfit-users mailing list
> Xgr...@li...
> https://lists.sourceforge.net/lists/listinfo/xgridfit-users
>

Re: [Xgridfit-users] Xgridfit 2.2

[Peter B. <b....@gm...>]

The attached patches fix the problem with the -S option. This option
does not work with the Saxon processor, which does not support the
EXSLT extension it depends on. I had disabled it when the output of an
Xgridfit run was Python, since I couldn't see the point--but that was
probably a mistake. And I may have unintentionally disabled it for
debug mode. It is still disabled for merge-mode, since the
organization of the output file is quite different in that mode, and
-S makes no sense at all there.

These fixes are in the CVS repository now.

On Tue, Mar 30, 2010 at 12:00 AM, Andrey V. Panov
<pa...@ca...> wrote:
> I have several issues with Xgridfit 2.2:
>
> 1. A "--root $(DESTDIR)/" argument should be added to "python setup.py install"
> command in Makefile.
>
> 2. The build of Istok font http://code.google.com/p/istok/ is broken with error
> messages:
>
> line 0: Extra element pre-program in interleave
> line 1039: Element xgridfit failed to validate content
> Error validating document
>
> 3. The build of Heuristica font http://code.google.com/p/evristika/
>
> xgridfit -m -p 25 -G no -i Heuristica-Regular_.sfd -o Heuristica-Regular.ttf -O
> Heuristica-Regular.py Heuristica-Regular.ed.xgf
>
> is broken with error message:
>
> execv() arg 2 must contain only strings
>
> 4. "-S ..." option to xgridfit does not work as expected even with xsltproc.
>
> --
> Andrey V. Panov
> http://canopus.iacp.dvo.ru/~panov/
>
> ------------------------------------------------------------------------------
> Download Intel&#174; Parallel Studio Eval
> Try the new software tools for yourself. Speed compiling, find bugs
> proactively, and fine-tune applications for parallel performance.
> See why Intel Parallel Studio got high marks during beta.
> http://p.sf.net/sfu/intel-sw-dev
> _______________________________________________
> Xgridfit-users mailing list
> Xgr...@li...
> https://lists.sourceforge.net/lists/listinfo/xgridfit-users
>

Re: [Xgridfit-users] Xgridfit 2.2

[Peter B. <b....@gm...>]

Andrey,

I've been working with Istok and have figured out how to generate the
missing *_acc* file. With that, it's clear that the problem is simply
that Xgridfit is now validating files by default. As I mentioned
earlier, this can easily be turned off. But it's just the equivalent
of the syntax checking that any compiler or interpreter does.

That said, though, I'll add that the current default choice of
validators is libxml2, and its error messages are sometimes cryptic or
even misleading. I suggest rnv instead. Install it, and then make it
the default:

xgfconfig --validators rnv

Then check over the Istok files:

for f in *.xgf; do xgridfit -x $f; done

One of the validation errors in Istok is just a matter of ordering
elements; perhaps Xgridfit is too picky there. But all the others I
came across certainly affect the quality of the output.

Peter

On Tue, Mar 30, 2010 at 12:00 AM, Andrey V. Panov
<pa...@ca...> wrote:
> I have several issues with Xgridfit 2.2:
>
> 1. A "--root $(DESTDIR)/" argument should be added to "python setup.py install"
> command in Makefile.
>
> 2. The build of Istok font http://code.google.com/p/istok/ is broken with error
> messages:
>
> line 0: Extra element pre-program in interleave
> line 1039: Element xgridfit failed to validate content
> Error validating document
>
> 3. The build of Heuristica font http://code.google.com/p/evristika/
>
> xgridfit -m -p 25 -G no -i Heuristica-Regular_.sfd -o Heuristica-Regular.ttf -O
> Heuristica-Regular.py Heuristica-Regular.ed.xgf
>
> is broken with error message:
>
> execv() arg 2 must contain only strings
>
> 4. "-S ..." option to xgridfit does not work as expected even with xsltproc.
>
> --
> Andrey V. Panov
> http://canopus.iacp.dvo.ru/~panov/
>
> ------------------------------------------------------------------------------
> Download Intel&#174; Parallel Studio Eval
> Try the new software tools for yourself. Speed compiling, find bugs
> proactively, and fine-tune applications for parallel performance.
> See why Intel Parallel Studio got high marks during beta.
> http://p.sf.net/sfu/intel-sw-dev
> _______________________________________________
> Xgridfit-users mailing list
> Xgr...@li...
> https://lists.sourceforge.net/lists/listinfo/xgridfit-users
>

Re: [Xgridfit-users] Xgridfit 2.2

[Andrey V. P. <pa...@ca...>]

Peter, 

validation with Xgridfit claims that <set-dropout-type value="5"/> is invalid. 
Newer TT instruction documentation from MS site describes such the modes with 
values greater than 2.

On 31 марта 2010, you wrote:
> I've been working with Istok and have figured out how to generate the
> missing *_acc* file. With that, it's clear that the problem is simply
> that Xgridfit is now validating files by default. As I mentioned
> earlier, this can easily be turned off. But it's just the equivalent
> of the syntax checking that any compiler or interpreter does.

By the way "inst_acc.py" produces Xgridfit code for instructing accented 
characters. This code uses external macros(functions), see "upr_functions.xgf" 
for instances.

> 
> That said, though, I'll add that the current default choice of
> validators is libxml2, and its error messages are sometimes cryptic or
> even misleading. I suggest rnv instead. Install it, and then make it
> the default:
> 
> xgfconfig --validators rnv
> 
> Then check over the Istok files:
> 
> for f in *.xgf; do xgridfit -x $f; done
> 
> One of the validation errors in Istok is just a matter of ordering
> elements; perhaps Xgridfit is too picky there. But all the others I
> came across certainly affect the quality of the output.

Thanks, it founds misprints, it is useful.

-- 
Andrey V. Panov
panov /@/ canopus.iacp.dvo.ru

Re: [Xgridfit-users] Xgridfit 2.2

[Peter B. <b....@gm...>]

2010/4/1 Andrey V. Panov <pa...@ca...>:
> Peter,
>
> validation with Xgridfit claims that <set-dropout-type value="5"/> is invalid.
> Newer TT instruction documentation from MS site describes such the modes with
> values greater than 2.
>

I set the max to 6 when it should have been 7. Changed in CVS.

Re: [Xgridfit-users] Xgridfit 2.2

[Peter B. <b....@gm...>]

2010/4/1 Andrey V. Panov <pa...@ca...>:
> Peter,
>
> validation with Xgridfit claims that <set-dropout-type value="5"/> is invalid.
> Newer TT instruction documentation from MS site describes such the modes with
> values greater than 2.

I've updated the schema (and my copy of the MS TrueType docs!): rnc
and rng versions attached.

>
> By the way "inst_acc.py" produces Xgridfit code for instructing accented
> characters. This code uses external macros(functions), see "upr_functions.xgf"
> for instances.

Very nice: thanks for pointing that out!

Re: [Xgridfit-users] Xgridfit 2.2

[Andrey V. P. <pa...@ca...>]

Peter,

I have another issue with validating xgf code. The validator claims that
<set-dropout-control threshold="255" flags="1"/> element is illegal. MS
documentation says that "A value of FF in bits 0-7 means invoke
dropout_control for all sizes." Threshold in Xgridfit is limited by 254.

On 3 апреля 2010, Peter Baker wrote:
> > validation with Xgridfit claims that <set-dropout-type value="5"/> is
> > invalid. Newer TT instruction documentation from MS site describes such
> > the modes with values greater than 2.
> 
> I've updated the schema (and my copy of the MS TrueType docs!): rnc
> and rng versions attached.


-- 
Andrey V. Panov
panov /@/ canopus.iacp.dvo.ru

Re: [Xgridfit-users] Xgridfit 2.2

[Peter B. <b....@gm...>]

Thanks -- fixed here and in CVS.

2010/4/5 Andrey V. Panov <pa...@ca...>:
> Peter,
>
> I have another issue with validating xgf code. The validator claims that
> <set-dropout-control threshold="255" flags="1"/> element is illegal. MS
> documentation says that "A value of FF in bits 0-7 means invoke
> dropout_control for all sizes." Threshold in Xgridfit is limited by 254.
>
> On 3 апреля 2010, Peter Baker wrote:
>> > validation with Xgridfit claims that <set-dropout-type value="5"/> is
>> > invalid. Newer TT instruction documentation from MS site describes such
>> > the modes with values greater than 2.
>>
>> I've updated the schema (and my copy of the MS TrueType docs!): rnc
>> and rng versions attached.
>
>
> --
> Andrey V. Panov
> panov /@/ canopus.iacp.dvo.ru
>
> ------------------------------------------------------------------------------
> Download Intel&#174; Parallel Studio Eval
> Try the new software tools for yourself. Speed compiling, find bugs
> proactively, and fine-tune applications for parallel performance.
> See why Intel Parallel Studio got high marks during beta.
> http://p.sf.net/sfu/intel-sw-dev
> _______________________________________________
> Xgridfit-users mailing list
> Xgr...@li...
> https://lists.sourceforge.net/lists/listinfo/xgridfit-users
>