[
https://issues.apache.org/jira/browse/PDFBOX-2050?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13987307#comment-13987307
]
Tilman Hausherr commented on PDFBOX-2050:
-----------------------------------------
I did the optimizations; The Flate filter is much faster, but the time spent in
that filter isn't very high anyway (in my test, 17ms before and 5ms after
optimizations - of a rendering of 3 PDF files that took one minute). But we're
done such optimizations before (PDFBOX-1006), so I'll do this too. Done in rev
1591811 for the trunk, and rev 1591812 for the 1.8 branch.
> Add predictor to LZW filter
> ---------------------------
>
> Key: PDFBOX-2050
> URL: https://issues.apache.org/jira/browse/PDFBOX-2050
> Project: PDFBox
> Issue Type: Bug
> Affects Versions: 1.8.4, 1.8.5, 2.0.0
> Reporter: Tilman Hausherr
> Assignee: Tilman Hausherr
> Priority: Minor
> Labels: LZW
>
> According to the PDF spec "LZW and Flate Predictor Functions", both can have
> post processing with predictors. It is implemented for Flate but not for LZW.
> I am adding this by using the existing code from the Flate filter, which I
> will be moving into a helper class.
> While looking at the Flate filter, I noticed that its footprint is higher
> than needed, i.e. there is too much buffering. I will test this separately
> and commit only if I can measure an improvement.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
