https://bugs.documentfoundation.org/show_bug.cgi?id=167321

--- Comment #29 from Dodo Ivanecky <[email protected]> ---
(In reply to Mike Kaganski from comment #28)
> (In reply to Dodo Ivanecky from comment #24)
> > practically: CSV = delimiter-separated values, **with the delimiter being
> > locale-dependent**.
> 
> NO.
> The piece "with the delimiter being locale-dependent" is wrong. It is not
> "locale-dependent". It depends on just about anything - any butterfly effect
> may prove relevant in deciding "which separators we chose in our bank" - it
> could be a moon phase of the date the decision was made.
> 
> We could *prioritize* some delimiters over others based on locale: let's
> say, that the data included both commas and semicolons, with ~same
> probability; then we could say "OK, the source locale chosen in the dialog
> uses comma as decimals, so assume semicolon in autodetection". Well, that
> *could* be a possible change; but that's not what happens here: your data
> has *no* other separators. So no matter if we prioritize one over another
> based on locales: in your data, when user asked "look at the data, and *try
> to guess* what column separators could be", the only possible answer is
> "comma", not "we decided that comma is not suitable, just because you are in
> Slovakia".

Taking locale into account is great. But since you are forced into the “CSV on
input” mode, it becomes more complicated. I am not using CSV — I am simply
taking 200 lines of decimal numbers (one per line) and pasting them in.

If you prioritize CSV detection first, then the problem is unsolvable. 3,142
becomes CSV 3 and 142, but for me, it’s π (PI)…

So I’m going to upgrade, and I hope that once I select “separated by,” it will
be remembered permanently!

-- 
You are receiving this mail because:
You are the assignee for the bug.

Reply via email to