Hi,
I am starting the work on issue 1910, which is about supporting non-ASCII
characters in URLs without the user having to encode and escape those
characters themselves. Also, I'm going to introduce auto-escaping of some
ASCII characters that unambigously can't be separators but that aren't
allowed in URLs (space and some others). This auto-escaping has to be done
in the user interface part (svn_opt_args_to_target_array).
IR support can be done at two levels:
- in the user interface (same as above)
- using IRI as the internal format, moving IRI->URI down to where IRIs
can't be used, such as in the HTTP protocol.
The first approach is simple and easy. It still uses URI (URL) as the
internal way of representing repository locations. But it forces us to do
URI->IRI in user messages, or live with encoded characters in messages,
which might be confusing to users.
The second approach is more risky, since it means changes on many more
places. It also introduces incompabilities, since all APIs that return
URLs now can return IRIs. We also have to decide on URLs in our data
files.
So, if no one has any good reasons, I'd like to use the first approach
above, keeping compability between 1.1 and 1.2. Any objections?
//Peter
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Tue Jul 20 21:19:50 2004