On 2014/07/19 03:10, Keith Medcalf wrote:
Mostly the problems experienced by people is that they make some home-
brew CSV importer that does not realise how to correctly read
output from a standards-based exporter such as Excel, and then try to
change things like separation or quoting methods to "fix" it
after the fact.
Which version of Excel produces standard-compliant CSV exports? As far as I have ever been able to tell,
Excel produces the most broken CSV files to be found anywhere. Of course, I haven't used any of the
"not-of-my-responsibility or under-my-control" (pronounced "cloud" by the marketroids)
versions, nor anything later than the 2010 versions. Most software designed to use Excel produced CSV files
(or to produce CSV files for use by Excel) have a "Use Excel Format" to handle Excel's
idiosyncrasies.
Please forgive me, that statement did not come out as intended - I actually have not thoroughly tested Excel, but I did test SQLite.
It was not intended as a claim for excel specifically - I simply meant standards compliant CSV exporters, such as "spreadsheet
programs" (Excel is just an easy name to grab). There is of course no guarantee any specific system adheres completely but they all
should, at the very least on the import side, be able to read ANY valid csv file. I have no idea whether Excel specifically
completely adheres or not, though in the same breath, I have no reason to think that it doesn't - if you have an example of 2007+
doing it wrong, I would very much be interested to hear about it.
As an aside - there is nothing preventing a spreadsheet, other system, or even something you may make to read/write more into a csv
formatted text than meets the eye, or specifically inundate it with system-specific little additions, this does not harm the
standard though, it simply adds functionality. A good example is adding \n\r to a field which is transparent to most
programs/importers in csv terms, but everyone else knows fairly well what that indicates. Whether this actually explains anything in
Excelian terms is unknown to me at this moment.
How excel interprets the data values once imported is a whole other matter that boggles the mind - such as insisting to convert any
imported series of digits into some dramatically weird INT or FLOAT value... such as telephone numbers. I would so prefer it
treating anything that is not already exactly a valid number, as text rather.
_______________________________________________
sqlite-users mailing list
[email protected]
http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
