On Tue, Sep 30, 2008 at 04:28:41PM -0700, Brock Pytlik wrote: > I had a question about _actions.c, lines 107 and 111. Will that method of > moving through a string work if it's a unicode string instead of a ascii > string? (And is that even a concern? I thought we wanted unicode support at > least for descriptions, but maybe that won't apply here.) It's probably > just a red herring, but I thought I'd at least raise the question.
So I took a quick look at what was going on in this area, and it looks like we're not regressing from the current behavior (which is what I expected). Basically, all the strings are passed through as-is, which means that anything encoded in any single-byte encoding that's a superset of ASCII or in UTF-8 will be just fine. I created a manifest with a single action, setting the description of the package to a string encoded in UTF-8. I published it and verified that the manifest in the repo was the same as what was passed from the client. I could install the package successfully, "pkg info" (and -r) both showed the description correctly, "pkg search" for that token found it and displayed it correctly, and the manifest and info pages on the BUI displayed correctly. I even fired up the GUI and made sure it displayed the string correctly. This is all for both the new code. So I'm pretty sure that we're handling our data reasonably well. I know we'll break down when trying other multibyte encodings, or any encoding that doesn't contain ASCII as a subset, but I'm not anxious to support those just yet. Danek _______________________________________________ pkg-discuss mailing list [email protected] http://mail.opensolaris.org/mailman/listinfo/pkg-discuss
