I'm kicking off the regression tests for 3.x. Nicholas, I merged TIKA-4303 and cherry-picked it back to 3.x. I hope that's ok.
On Fri, Jan 24, 2025 at 2:25 AM Tilman Hausherr <thaush...@t-online.de> wrote: > Hi, > > No opinion re release schedule but a comment on the PDFBox update: > > tl;dr: ignore the PDF differences this time. > > The new version includes the /ActualText support: > https://issues.apache.org/jira/browse/PDFBOX-5868 > > It is always enabled. In most cases the extraction is better. But > sometimes content is lost because the feature is used for obfuscation > (see example in the issue above). > > Another major change is the detection of the space width: > https://issues.apache.org/jira/browse/PDFBOX-5920 > It has been improved, however this will result in many differences with > angled texts if angle detection isn't enabled. Some scientific texts > with superscript prefix will also look different, "1 Coupled" will > extract as "1Coupled". This is because these fonts don't have a space > and the fallback we are using sucks. > > Tilman > > On 16.01.2025 14:20, Tim Allison wrote: > > Sorry, on second thought, a small tweak: > > > > I propose that we release 3.1.0 after PDFBox 3.x is released. I further > > propose that we make a 2.9.3 release at some point after the 3.1.0 > release > > IF we get requests for a 2.x release...otherwise we'll do a final 2.x EOL > > release in April, 2025. > > > > On Thu, Jan 16, 2025 at 8:15 AM Tim Allison <talli...@apache.org> wrote: > > > >> All, > >> It has been a while since we last released 2.x (April 2024) and 3.x > >> (October 2024). We've had a number of dependency updates. PDFBox is on > the > >> cusp of a new 3.x release. > >> I propose that we release 3.1.0 after PDFBox 3.x is released and > that we > >> make a 2.9.3 release the following week. > >> WDYT? > >> > >> Best, > >> > >> Tim > >> > >
