> Reading the RFC only proves my point. SQLite v3 is older than that > doc, and it pretty much admits the cat was out of the bag a long time > ago. There are a ton of optional and might/may/could sections that > event the format they define has a lot of holes in it (i.e. headers, > or no headers?).
I read the RFC last night and even took a stab at adding it to the .import method in the sqlite command line. But pretty quickly holes appeared. For example, the RFC says that everything between the commas is supposed to be part of the field, including white space. Normally I trim the white space unless it's quoted. Still, the RFC does not address how to handle rows like this: 1234,abc"123",abc 1235,""123,abc What are you supposed to do with those? It is not clear. Also, are you supposed to strip the quotes upon consuming the field? Are you supposed to un-escape escaped quotes? "1234" -> 1234 or "1234" ? "15""" -> 15" or 15"" or "15""" or "15"" ? Seems to me if you strip quotes, you have to un-escape any escaped quotes in the field. If you don't strip quotes, then you can't un-escape anything in the field because it leads to bazaar edge cases. Then there is the matter of white space outside the quotes. The RFC seems silent on all these issues, though the ABNF grammar implies that white space outside quotes is not tolerated, which could lead to considerable user surprise. RW Ron Wilson, Engineering Project Lead (o) 434.455.6453, (m) 434.851.1612, www.harris.com HARRIS CORPORATION | RF Communications Division assuredcommunications(tm) _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
