Hi all,
I notice that many, if not all of the archived repos that we provide contain BOTH the original *.jar artifacts as well as the packed *.jar.pack.gz artifacts.
This seems like a big waste of disk space and download bandwidth to me.
There was some case for the unpacked *.jar artifacts in the past, when Java 1.4 and below was still widespread and thus the unpack utility was not universally available. But given that more and more projects now require Java 1.5, keeping
the unpacked *.jar artifacts in the archives seems unnecessary and wasteful. A similar argument holds for the actual online accessible repos.
To give you a rough idea of how much could be saved, I looked at a local copy of the Helios SR2 repo (not claiming that removing all unpacked Jar’s from Helios is a viable option yet):
helios> du -k 201102250900
1008535 201102250900
helios> find . -name '*.jar.pack.gz' -print | sed -e 's,\.pack\.gz,,g' | xargs rm
helios> du -k 201102250900
507807 201102250900
As you see, disk usage goes down from 1 Gig to 500 Meg. Multiply this by all our mirrors plus the bandwidth taken for downloads… and consider that the argument applies for Project’s individual repos, multiple versions of downloadable archived
repos, … and see how this could be related to
https://bugs.eclipse.org/bugs/show_bug.cgi?id=335809 .
Thoughts, comments anyone ?
Any good reason for not cleaning the upacked Jar’s ?
Thanks,
Martin
--
Martin Oberhuber, SMTS / Product Architect – Development Tools,
Wind River
direct +43.662.457915.85 fax +43.662.457915.6