In fact, we use true type font and we added in PDFRenderer an hashMap which has the mapping between the characters and their unicode, so when the hashMap contains the character, we extract the corresponding unicode to render, else we let the conversion method doing her job.
Hope this will help

Renan

Victor Mote a écrit:
Renan Collin wrote:

  
there could be a good performance improvement by customizing 
FOP, it encodes every single character it found in the XML in 
PDF without testing if it has already been encoded, so upon a 
large large amount of data, it could be a good solution.
By doing this FOP is 3 times more rapid on 5 to 10 document pages.
    

Hi Renan:

Thanks for your comment. I am working on a font library (derived from FOP
0.20.5) that is being considered for possible integration with FOP. Your
comment sounds familiar, but I wanted to make sure I understood exactly what
you are saying here. About a week ago, I completed a round of changes to
this code that put encoding, character sets, and kerning into more efficient
data structures. I have a similar upcoming project that is intended to do
the same for the font subsetting logic. IIRC, FOP 0.20.5 uses two HashMap
instances to manage the re-encoded subset values, and I am sure we can
improve that. My current plan is to use parallel arrays to track the
relationship.

So my questions: Is your comment directed at subsetted TrueType fonts? Would
you please elaborate more on both the problem that you discovered and the
solution that you have developed? I would very much like to make sure that
we don't miss an opportunity for a 3x performance improvement, and would
appreciate any additional information you can provide.

Victor Mote


---------------------------------------------------------------------
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]

Reply via email to