I realize now the difference is in hyphenation. Our application uses
hyphenation but it's not kicking in from my command-line fop invocation
though the fop-hyph.jar is in the classpath. Can this be done in the
-cfg config.xml? i.e mimick the call to Factory.setHyhphenBaseURL().
rjs
On 07/06/2012 02:50 PM, Rob Sargent wrote:
I'm using FOP-1.0, java version "1.6.0_31" on ubuntu-12. I'm trying
to develop a test fo which shows what I believe is a layout failure in
FOP. In certain circumstance, related to layout choices, a page
begins with an unwanted indentation and a tiny amount of text lost ( a
dozen characters at most, a word or two ).
Unfortunately my command-line run of fop, local build also installed
in production, does not use the same fonts (at least not the same
size) as our running application. I am running this command line:
fop1 -c ./cfg.xml -fo fop.xml -pdf t.pdf > fop.log 2>&1
and in the cfg.xml I specify the base and font-base to that used in
production. I'm confident the latter is taking affect since with out
it I get complaints about one of the fonts called out in the fo as
missing. Adding the cfg.xml, otherwise identical to the production
config file, resolved this issue (which is actually an alias of
Helvetica to FreeSans).
The fonts used are TimesNewRomanPSMT (main text), Optima-Oblique
(headings), Helvetica-cum-FreeSans for special characters,
GillSansMTPro (bulleted text) and Times-BoldSC (small-caps, denotations).
In the log of the run I see Lookups for only TimesNewRomanPSMT and
Optima-Oblique.
2012-07-06 13:43:17,147 [main] DEBUG org.apache.fop.fonts.FontInfo
- Font lookup: TimesNewRomanPSMT normal 400
2012-07-06 13:43:17,181 [main] DEBUG org.apache.fop.fonts.FontInfo
- Font lookup: Optima-Oblique normal 400
Does this suggest that it is finding system supplied definitions of
those font.
My eye tells me the main font is TimesRoman but I'm no typographer by
any means and it could be a variant. The other non-looked-up fonts are
also in play and the block GillSans measures identically in
application and command-line pdf.
With exact font (metric) replication I don't think I can make the test
case and without a fix or work-around we're at high risk of making
fools of ourselves publicly. (A truth to be sure, but the public
display isn't needed).
Any thoughts on correcting this are greatly appreciated.
rjs
