build script should support building without an internet connection
-------------------------------------------------------------------
Key: PDFBOX-587
URL: https://issues.apache.org/jira/browse/PDFBOX-587
Project: PDFBox
Issue Type: Bug
Affects Versions: 0.8.0-incubator
Environment: development
Reporter: Mel Martinez
The build scripts for PDFBox currently depend on having an internet connection
in order to build. This should be fixed so that the distribution can be built
in cases where no internet connectivity is available.
The purpose of the connectivity is to download various dependencies. For the
most part that is fine. The scripts check for the existence of needed files
and if not found retrieve them.
So for most cases, simply manually pre-populating the /externals, /download and
/tools folders with the needed files works just fine.
The one place where the script still breaks is the dependency on the
glyphlist.txt file which it looks for in the /target/classes/Resources folder.
Since the initial 'clean' target at the start of the 'dist' build wipes out
the 'target' folder, then that dependency never 'exists' and the build script
will ALWAYS try to download that file and fail if there is no internet
connection.
The solution is fairly simple: That file should be downloaded first to the
/downloads directory and then copied to the /target/classes/Resources
directory. That allows you to manually seed the file into the /downloads
directory and build without a connection.
Fixing this in the build.xml script is very straightforward:
line 115 should change
from: <property name="adobefiles.glyph.txt"
value="${dest.dir}/Resources/glyphlist.txt"/>
to: <property name="adobefiles.glyph.txt"
value="${download.dir}/glyphlist.txt"/>
line 193 should have the following line inserted
<copy todir="${dest.dir}/Resources/" file="${adobefiles.glyph.txt}"
overwrite="true"/>
That's it!
To really do this nicely, it would be nice if all the downloads were first
placed into /downloads and then copied to /externals, /tools, etc. , and that
all such downloaded files were concisely documented or at least listed out in
one place in the script so that one could know what to go retrieve manually.
But just the above fix would be a major improvement.
I know we are moving PDFBox to a Maven build, but unfortunately I still need to
be able to build within a fairly restrictive environment.
Thanks.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
