The upshot on posting from a Mac browser
I traded a couple of emails about this with Jake Savin, but his attention is needed on something more pressing. I just want to make notes while this is still almost fresh in my mind. Caveat: much of these conclusions are my best guesses, from observing the results of my browser tests. If you know better, let me know.
All of the Mac browsers that I tested send posted data in iso–8859–1 encoding, a.k.a. Latin–1. Iso–8859–1 is an 8-bit encoding scheme, which includes 256 characters.
The curly quotes, n-dash, m-dash and other characters—that are very desirable from the designer’s point of view—are not within the iso–8859–1 character set.
Mozilla sends the characters using UTF–8 Unicode, a modern Internet standard. All other Mac web browsers send these using Windows CP–1252 encoding, which is similar to iso–8859–1, but adds these characters into an unused block of codes. I was surprised to find that even Mac-only browsers like OmniWeb and iCab do this.
A Manila or Radio server running on a Mac takes the incoming text and does an iso to Mac-Roman conversion, so that the text is stored internally in a native format. Unfortunately, the iso-Mac conversion bungs up the Windows characters.
The practically-minded solution would be to extend the iso–Mac converter (called string.latinToMac) so it understood this, but that stinks of embrace-and-extend. If I have time, I’ll use some string and rubber bands to test this on my own installation of Frontier and the Radio demo.
In the mean time, I’ll keep typing “’” for apostrophes. 12:31:23 AM
|