Re: multiple plot lines with plot '-'

[Daniel J S. <dan...@ie...>]

On 05/16/2012 11:50 AM, Allin Cottrell wrote:
> On Wed, 16 May 2012, Daniel J Sebald wrote:
>
>> Perhaps there should be a mode switch, "use the same data" or "don't
>> reread". But I'm not sure how crucial that is.
>
> Just a notion, but what if gnuplot's syntax permitted defining a matrix
> (either "inline" with some nice, simple syntax or by reading from a
> file), and you could then say something like
>
> plot matrix "foo" using ...
>
> as many times as you wanted, using whichever columns?

Well, that's a paradigm change as far as programming because the whole 
plot command would need to be processed to figure out what columns are 
required for _all_ plots not just the current one.  That also expands 
the number of possible storage elements creates difficult programming in 
this design.

Currently, the data file could be (exaggerating) 1000 columns wide but 
for any given plot gnuplot selects and stores only the few columns of 
data it needs for individual plots (say column 1, column 50 and column 
51 on the first plot).  It does the same for the next plot, but it 
rereads the file and might select a different few columns (say column 1, 
column 120, column 121).

Attempting an explanation, gnuplot has been programmed in such a way as 
to not "look ahead" at what is to come.  Changing to a system that needs 
to "look ahead" for plots that are to come really creates challenges 
internally.  gnuplot started in an era in which memory wasn't so readily 
available, but at the same time I think it should probably remain close 
to the parsimonious paradigm as possible.  In my opinion, software 
generally has become a little too bloated.  I did write a patch a while 
back that stored the raw data, sort of bucking that paradigm, but that 
was storing just the columns of data an individual plot uses, not the 
whole possible data file, i.e., no "look ahead" but instead "store what 
has been used".

Dan