[Kst] Re: branches/work/kst/portto4/kst/src/datasources/ascii

[Peter Kümmel]

On 24.01.2011 14:15, Brisset, Nicolas wrote:
> Hi Peter,
> 
>> SVN commit 1216727 by kuemmel:
>>
>> AsciiSource: Add option to speedup by a priori information of data.
>> When columns have ALWAYS the same with skip the searching for a column
>> start after it was found the first time.
> Could you elaborate on that a bit? It sounds like a good idea, but I
> don't have time to look at the code in detail and I don't exactly
> understand what should be constant to be able to use this option:
> - the type of comment delimiter
> - the type of column delimiter 
> - the number of characters per column (i.e. columns should be
> fixed-width)
> - the position of the first character of the number to read (including
> sign), i.e. what happens if the numbers are always somewhere within a
> column of a given width, but not always left-aligned on the column
> boundary?
> - ...
> 
> Fixed-width columns used to be optimized in the past, and that's why
> there is this option in the current GUI. As that already exists, you
> probably mean something else. But again, this is not completely clear to
> me.

Fixed-width columns are optimized, column width == 5
 1234 1234 1234 1234;
-1234-1234 1234 1234;
-1234 1234 1234 1234;
+1234 1234-1234-1234;

but we could also optimize for different column width are constant:
 1234; 123456789; 12; 1234444;
 1234;-123456789; 12;+1234444;

only the place of the ; has to be the same in each row.

And I've seen no problems when the number string starts with a whitespace.

Peter