#227645 Document how to plot compressed file

Package:
gnuplot-doc
Source:
gnuplot
Submitter:
Tim Connors
Date:
2018-11-26 14:36:03 UTC
Severity:
wishlist
#227645#5
Date:
2004-01-14 03:20:11 UTC
From:
To:
I have some extremely large data files that I compress and decompress
on the fly with all of my analysis programs. There is one spanner that
is thrown in the works though -- and that is that gnuplot doesn't
support on the fly decompression using e.g. zlib.

Maybe use a heuristic like file to work out what kind of compressed
datafile, or support an option to tell gnuplot that the file is a gzip
or bzip2 or compressed file, or just work on the filename (the latter
has its flaws, because there are sometimes reasons why filenames are
not given the .gz|.bz2 extension)
#227645#10
Date:
2004-01-20 03:01:04 UTC
From:
To:
Woops, I forgot about that syntax. Yes, thankyou - it does work.

CCing debbugs for the sake of future google users.

I wonder if there could be an extra help file "filters" in gnuplot, since
it wasn't obvious to me to goto the "special filenames" section?
#227645#15
Date:
2010-09-14 16:23:53 UTC
From:
To:
tags 227645 upstream
retitle 227645 Document how to plot compressed file
assign 227645 gnuplot-doc
forwarded 227645
https://sourceforge.net/tracker/?func=detail&aid=3066055&group_id=2055&atid=102055
thanks
#227645#28
Date:
2010-11-26 17:18:09 UTC
From:
To:
forwarded 227645 https://sourceforge.net/tracker/?func=detail&aid=3066055&group_id=2055&atid=102055
tag  227645 + patch
thanks

Upstream will not change his documentation, so debian should locally patch
#227645#39
Date:
2018-11-26 14:32:46 UTC
From:
To:
It isn't really a defect in gnuplot, because it can be done and is documented.

plot "< xyz" ...

(piped-data) is extremally powerful, often used with sed, grep, awk, cat/zcat or other tools
to do arbitrary preprocessing that can't be done with input spec or 'using'.

I agree, that documentation could show more examples, because
it is so powerful and underutilized.

Automatic heuristic to discover some magic headers for gz/bz4/xz/lz4/lzo
plus heuristic to look at the extension of the filename, could be useful
useful tho, especially for new users, or:

plot "file.gz" decompress=auto ...

to not even use heuristic. The problem with heuristic would be that it
will not work for tar.gz files, or zip files, as these can contain more than
one file, but initial support could simply treat them as separate datasets
or just use first one and error out if there is more than one.

Similar to plot "xyz.png" binary filetype=png can do read arrays.
In fact for some simple formats simply plot "..." binary filetype=gzip could
do the trick.

In the mean time it would be nice to document this upstream as one of example.


Best regards,
Witold