Tim Allison created TIKA-5005:
---------------------------------
Summary: replace ttf-mscorefonts-installer with metric-compatible
fonts (Liberation, Carlito, Caladea)
Key: TIKA-5005
URL: https://issues.apache.org/jira/browse/TIKA-5005
Project: Tika
Issue Type: Task
Reporter: Tim Allison
I was recently annoyed that we had a failed build because of sourceforge:
[https://github.com/apache/tika/actions/runs/40508823275] .
Then I looked a bit more at licensing, and I think we should try to drop
mscorefonts.
:robot:'s input:
{noformat}
The -full and tika-grpc images install ttf-mscorefonts-installer, which
downloads the fonts from SourceForge at build time, once per architecture. A
mirror outage failed the arm64 snapshot push (run 40508823275). Earlier, the
same package forced linux/arm/v7 off the platform list. Shipping an image with
the fonts already installed also amounts to redistributing them under the MS
EULA. Proposal: remove it (plus cabextract/wget if nothing else needs them) and
add fonts-crosextra-carlito and fonts-crosextra-caladea. Before deciding, scan
the PDF corpus for non-embedded fonts whose BaseFont is Verdana, Georgia,
Trebuchet, Comic Sans, Impact, Webdings or Andale Mono, to bound how many
documents would be affected.
{noformat}
Let me know what you think?
--
This message was sent by Atlassian Jira
(v10.3.26#10030026)
