Hi Hervé,
So I've looked the patch over quite a bit and I have some
questions/comments. In general stuff looks good, I'm sure
people dealing with WMF files will appreciate the significant
improvements there, unfortunately I know nothing about WMF so
I don't have much to comment on that work.
XmlWriter changes right now I'm not going to accept due to some issues:
1) Your updates removed the use of bulk writes which was
specifically added due to performance issues.
2) The change to encode all chars >= 0x80 as hex is not friendly
for foreign language users (they might view the document in an
editor that really displays those things). This is one of the
intents of encoding to handle. At most this might be made a option
for when writing the document.
SVGGraphics2D
Line 223 in patch:
What's with the null paint stuff? It looks like it isn't used.
Lines 272 & 292 in patch:
I'm not sure that the 'solution' of setting stroke to the
'textColor' is a good one - most text doesn't want to be stroked
at all. Can you describe exactly what the problem is?
Should we be setting stroke to 'none'?
Line 289:
In what case are we missing setting the font-size correctly?
Is this for cases where you are using 'getRoot' repeatedly?
If so then much more needs to be done I think...
Line 305:
You deleted the restoration of the regular transform after a
supplemental transform was needed to draw text (needed for very
small fonts).
Line 329:
While I appreciate the very good attempt at handling of attributed
Character iterators, I am concerned that the code only considers
the attributes on the first char. These attributes can change
on a per char basis - requiring the creation of tspans. Perhaps
check code can be added to see if the attributes are constant
(see 'ACI.getRunLimit(Set attributes)' and use the improved code
in that case and use the outline path it in the more complex cases?
Otherwise I'm afraid people will see
regressions (text will be text but improperly styled).
This is a good start towards generating the tspans (essentially the
same code needs to be done for each tspan).
Line 430:
I'm not thrilled with the mapping of Justifications. Once again
I would lean towards mapping zero to 'start', one to 'end'
.5 to middle, any other value can either be done with 'dx' or
simply stroke it. This should handle the most common cases nicely
and not cause regressions for the other cases.
Line 3936:
.nbattrs (Net beans something file) in wmf/tosvg needed?
Really annoy bits: ;)
You seem to have been a bit careless with the Copyright lines.
Most of the new classes say 2001, in a few cases you deleted
existing years in a Copyright.
Finally:
Can I add the 'test' classes to Batik? They have the Apache License
so I assume so but I just wanted to make sure. I will probably
add the WMF viewer to the 'contrib' directory and try and figure
out how to merge the test transcoder into the Batik test framework.
Thanks again this looks like a big improvement in the WMF stuff,
as well as some potentially very useful improvements to the
SVGGraphics2D.
----
Thomas DeWeese wrote:
Hi Hervé,
Hervé Girod wrote:
I have enhanced / hacked the actual transcoder code (as of version
1.6) and added / corrected the following functionalities :
[...]
OK, it is done (much more mork than I expected, since I tried to
"beautify" my code a little). Also I added image / texture handling.
Thanks I'll look at integrating this soon.
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
