I think you could significantly improve the compression on the odt docs.

users
#1

I think you could significantly improve the compression on the odt docs.

tar -c dir | xz -4 -e >> file.tar.xz

16,760 tar.xz
21,919 odt

76.46334230576213%

If you were really crazy you could use a custom filter chain to knock off
an additional 0.4927232081755619%.

xz -k -C sha256 --lzma2=dict=64MiB,lc=4,lp=0,pb=0,mf=bt4,mode=normal,\
nice=273,depth=500 file.tar

75.97061909758656%

16,652 tar.xz

And it only takes my system (64bit 3.2 GHz,)

real 0m0.399s
user 0m0.248s
sys 0m0.148s

to complete the above custom filter chain task so it can't be that hard
for an older system to do.

The file I tested this on was a very old file from before I knew much
about computers. I created it because intranet explioter can't manage to save
wiki pages and I was most interested in saving the file on how to create a level.
So it should be freely reditributable and I include it for the interseted.

It's decompressed size is

213,906 bytes

Yielding

10.247024393892644% odt

and

7.7847278711209595% tar.xz

for the custom filter chain.

I think a savings of about 25% is worth looking into.

#2

Hi :slight_smile:
It might be good to contact the developers about that. Perhaps post a
bug-report and then use the drop-downs to turn into a "feature
request" for someone keen to jump on. 76% is huge!!
Regards from
Tom :slight_smile:

#3

Hi Frank,

frank ernest schrieb:

I think you could significantly improve the compression on the odt docs.

tar -c dir | xz -4 -e >> file.tar.xz

16,760 tar.xz
21,919 odt

76.46334230576213%

there is no choice, how to compress the package, because it is defined in the standard, see http://docs.oasis-open.org/office/v1.2/os/OpenDocument-v1.2-os-part3.html#__RefHeading__752789_826425813

Kind regards
Regina

#4

​Agreed.
It *is* possible to achieve better compression than ZIP (we knew how to do
that ​for quite some time :wink: ), but to do so we would need a new iteration
of the OpenDocument standard. While it's not totally out of question, I
have the feeling that increasing fragmentation just to save space on office
files is not a good trade-off. Remember that ODT files are used by many
pieces of software, not only LibreOffice.
OpenOffice of course, but some part of google export, some less-known
office suite, various custom software that output their reports as odt
files... even mso got to implement it. I doubt a new version of the
standard would be a good thing while it is still functionnal.

#5

Hi :slight_smile:
Ahh, ok :slight_smile: That makes sense now.

It is, of course, possible to put several Odt documents inside another
zipped file and that should benefit from the higher compression rate.

I guess the "feature request" would be passed on to the OASIS
committee but it sounds like they have probably already weighed-up the
pros-and-cons
Regards from
Tom :slight_smile: