fprintf for blocks/vectors: why the \n?

fprintf for blocks/vectors: why the \n?

[Fleur Kelpin]

Hi all,

If the automatic \n in block/fprintf_source.c were to go, people could
provide the format "%g\n" or "%g\t" or anything they like.

This makes it easier to send the output of a time series of vectors to
e.g. gnuplot because one can also print to columns then, instead of
the now compulsory elements below each other.

Or, if backwards compatibility is a concern, maybe provide this in a
second function with slightly different name?

The way things are now, I find I am writing my own for loops for printing,
whereas gsl tries to provide this functionality.

Greetings,
Fleur Kelpin

fprintf for blocks/vectors: why the \n?

[Fleur Kelpin]

Hi all,

If the automatic \n in block/fprintf_source.c were to go, people could
provide the format "%g\n" or "%g\t" or anything they like.

This makes it easier to send the output of a time series of vectors to
e.g. gnuplot because one can also print to columns then, instead of
the now compulsory elements below each other.

Or, if backwards compatibility is a concern, maybe provide this in a
second function with slightly different name?

The way things are now, I find I am writing my own for loops for printing,
whereas gsl tries to provide this functionality.

Greetings,
Fleur Kelpin

Re: fprintf for blocks/vectors: why the \n?

[Brian Gough]

Fleur Kelpin writes:
 > If the automatic \n in block/fprintf_source.c were to go, people could
 > provide the format "%g\n" or "%g\t" or anything they like.

Hi,

That function is just intended as an ascii version of the binary write
function, for portability purposes.  Hence it dumps the list of
numbers to the stream without any structure.

 > The way things are now, I find I am writing my own for loops for printing,
 > whereas gsl tries to provide this functionality.

Yes, to actually display a vector or matrix you'll need to write your
own (little) display routine which prints them in your favorite
format.  Then you can call one of those routines whenever you need to
display a vector or matrix.

If somebody writes some good routines they can go into the library.
Ideally they should be able to display in all the standard formats
like Fortran, Octave, HTML, TeX, troff, etc using an argument to
select between them. 

regards

Brian

Re: fprintf for blocks/vectors: why the \n?

[Brian Gough]

Fleur Kelpin writes:
 > If the automatic \n in block/fprintf_source.c were to go, people could
 > provide the format "%g\n" or "%g\t" or anything they like.

Hi,

That function is just intended as an ascii version of the binary write
function, for portability purposes.  Hence it dumps the list of
numbers to the stream without any structure.

 > The way things are now, I find I am writing my own for loops for printing,
 > whereas gsl tries to provide this functionality.

Yes, to actually display a vector or matrix you'll need to write your
own (little) display routine which prints them in your favorite
format.  Then you can call one of those routines whenever you need to
display a vector or matrix.

If somebody writes some good routines they can go into the library.
Ideally they should be able to display in all the standard formats
like Fortran, Octave, HTML, TeX, troff, etc using an argument to
select between them. 

regards

Brian

Re: fprintf for blocks/vectors: why the \n?

[Gerard Jungman]

Fleur Kelpin wrote:
>
> If the automatic \n in block/fprintf_source.c were to go, people could
> provide the format "%g\n" or "%g\t" or anything they like.
> 
> This makes it easier to send the output of a time series of vectors to
> e.g. gnuplot because one can also print to columns then, instead of
> the now compulsory elements below each other.

I never noticed this. I agree; the way it is now is not right.


> Or, if backwards compatibility is a concern, maybe provide this in a
> second function with slightly different name?

Brian, what do you think?


-- 
G. Jungman

Re: fprintf for blocks/vectors: why the \n?

[Gerard Jungman]

Fleur Kelpin wrote:
>
> If the automatic \n in block/fprintf_source.c were to go, people could
> provide the format "%g\n" or "%g\t" or anything they like.
> 
> This makes it easier to send the output of a time series of vectors to
> e.g. gnuplot because one can also print to columns then, instead of
> the now compulsory elements below each other.

I never noticed this. I agree; the way it is now is not right.


> Or, if backwards compatibility is a concern, maybe provide this in a
> second function with slightly different name?

Brian, what do you think?


-- 
G. Jungman