Alvaro Herrera <alvherre@commandprompt.com> writes:
> Excerpts from Leslie Satenstein's message of dom dic 19 23:39:38 -0300 2010:
>> The insert examples in section 2.4 do not function if a cut and paste from
>> pdf to psql is done
>>
>> This is the problem
>> INSERT INTO weather VALUES (âSan Franciscoâ, 46, 50, 0.25,
>> â1994-11-27â);
>> The fields city and date should have ' as delimiter, not
>> â
>> Corrected is the following that works. Error messages are not clear. It
>> would be great if psql indicated invalid character as error type.
> The problem is that â seems to be a valid character for identifiers:
Yeah, the parser will take most non-ASCII UTF-8 characters as being
identifier characters. But the real question is why Leslie is getting
â and not plain ' when he copies and pastes.
What I can see from here:
* the character in the SGML source is definitely plain '
* what is in the PDF file displays like a fancy curly quote
* copying and pasting it, however, gives plain ' for me
The PDF file I used was built by me on Fedora 13; it may or may not
exactly match what Devrim distributes. I used Preview on OS X to
display the PDF.
So what it looks like from here is that platform-specific differences in
the copy/paste process may be at issue.
Whatever the true story, I don't think there's a darn thing we can do
about it. The characters are correct in the SGML source, and we are not
in a position to fix whatever subtle bugs may or may not be present in
the PDF generation process.
regards, tom lane