On 25.01.2019 14:08, Robert Elz wrote: > Date: Fri, 25 Jan 2019 12:38:00 +0100 > From: Kamil Rytarowski <n54%gmx.com@localhost> > Message-ID: <9790a289-8b3e-e57a-e4d8-c021aa7b1b38%gmx.com@localhost> > > | I think it's better to keep an uniform way of handling separator in > | basic command line tools and restrict locales to translations only. > > Once again, if this is to change, it needs to be discussed and > agreed on an appropriate list, in messages that are obvious enough > that anyone interested is likely to read them. > > As my (currently uncommited) code is, making the change would > be trivial - just deleting two lines and changing the indentation on > the one that follows (and updating a few comments.) (Plus one of > the new ATF tests would need changing.) > > Or, there could be an option, which could be used in cases where > the arg has come from user input, rather than being coded into a > script (the program being run cannot tell from where its command > line args originate - it would need to be told) and only do locale > specific parsing when expressly told to do so (in which case when > so instructed it would only allow the correct locale specific value). > That would not be much harder (and very little extra code - though > it would need documenting.) > > | My native locale uses ',' for separator... > > Yes, most of Europe does I believe. Someone (from Europe) once > tried telling me that it was "everywhere" except English speaking > countries. But that was a very Euro-centric viewpoint (only Europe > (plus North America) counted as part of "everywhere") and is > certainly not correct in general. > > It isn't really relevant, but it is worth pointing out, that technically > in English the radix char is not '.' either, that's just what we use > in computing and typing (which these days means almost everywhere). > The actual character is slightly bigger than a period, and positioned > about half way up the height of a normal lower case character (the > digits should have both ascenders and descenders, and not all be > equal height ... but that is even less common these days than a > correctly implemented decimal point - which would appear about > half way up the loop in a 6 or 9, which should both have their > loops at the same horizontal position.) > I think that it's similar in European countries like Poland. The separator most likely originated from handwritten practice of a symbol between , and . that in calligraphy resulted in a differences approximated to either dot or comma depending on the countries (/fonts). Nowadays using , is software is in my perspective appropriate almost only in Microsoft Office and its clones and other software oriented towards producing printed text with a good style according to local formal practice (for rather non-technical people and non-technical purposes), while all/most technical tools and books use dot. Comma is more useful to separate different numbers in sets, vectors or similar. The purist approach of handling the separator in command line tools (or math books) accordingly to locale is in my opinion hypercorrectness. > kre >
Attachment:
signature.asc
Description: OpenPGP digital signature