[Seaside] Seaside 2.7

[Philippe Marschall]

2007/2/28, Koji Yokokawa <koubo2006 at yengawa.jpn.org>:
>
> On Wed, 28 Feb 2007 09:17:01 +0100
> "Philippe Marschall" <philippe.marschall at gmail.com> wrote:
>
> > 2007/2/28, Koji Yokokawa <koubo2006 at yengawa.jpn.org>:
> > >
> > > On Tue, 27 Feb 2007 15:36:19 +0100
> > > "Philippe Marschall" <philippe.marschall at gmail.com> wrote:
> > >
> > > > 2007/2/27, Masashi UMEZAWA <masashi.umezawa at gmail.com>:
> > > > > Hi,
> > > > >
> > > > > 2007/2/25, Koji Yokokawa <koubo2006 at yengawa.jpn.org>:
> > > > > > I have some minor changes to fix troubles about UTF8 and WideString.
> > > > > > (I don't have tests but I checked in Japanese)
> > > > > >
> > > > >
> > > > > Unfortunately, these patches are only for Squeak. Since Seaside should
> > > > > be easily ported to VW (or even Dolphin?), I think there need to be
> > > > > more indirections. (Especially the patches for WAUrl).
> > > >
> > > > Not so much. We already send #escapeForHTTP in WAUrl so sending it at
> > > > a second place won't break anything that is not already broken. It
> > > > won't do utf-8 encoding in Squeak 3.8 but fixes functionality and
> > > > breaks done.
> > >
> > > And, that WAUrl patch for Squeak3.8 is a temporary work-around to avoid
> > > exception. I think String>>escapedForHTTP should convert the WideString
> > > to utf-8 at the sametime, like the implementation in Squak3.9.
> > >
> > > > The sever adapters are platform specific anyway (they are located in
> > > > 'Seaside-Platform' too) so this is no problem as well.
> > > >
> > > > It added the patches in a slightly changed form (there was a bug when
> > > > copying urls) to Seaside 2.8.
> > >
> > > Thanks.
> > >
> > > > One thing to always remember it that some people do not want
> > > > WideStrings and prefer ByteStrings in utf-8 instead.
> > >
> > > Well, but some users will post wide-strings easily, they don't care
> > > about the developer's intention. Then some application will be crushed.
> > > I noticed that Pier is broken by a post in Japanese.
> >
> > Then this is a bug in your Pier setup. Pier see below how Pier
> > supports posts in Korean
> > http://www.lukas-renggli.ch/blog/studenckifestwal?command=PRViewCommand&view=PBPostHtmlView
> >
> > This setup uses WAKom and therefore ByteStrings. ByteStrings don't
> > mean you can't support non-latin1 languages.
>
> Good!
> Sorry, it's my misunderstanding. Pier is NOT broken by Japanese text too.
>
> I agree it's enough to display it in a web-browser.
> However, it is not the point I'm afraid. I mean the non-latin1 text is
> broken as a WideString in Squeak. The text cannot be saw and handled
> crrectly in the Squeak side, not in the web-browsers.

That's true. But Pier is (mostly) a web application. It (in general)
doesn't display morphs and stuff so this is less of an issue. Even if
it did display morphs it whould probably be a pain to getting all the
characters displayed correctly. I can tell you it is a pain to paste
non-latin1 (some times even non-ascii) text into Squeak.

> So that, you
> cannot edit, even search, the text from Squeak. And, I think, it will be
> cause of many bugs.

Kinda depends what you are doing. Most web applications I have seen
just consume strings and ouput them again unchanged. You only have to
pay attention when you are creating substrings (the first n
characters). Naive text searching might return a few false positives.
Sure, if you are building a suffix tree or boyer moore text search and
stuff then you really have to pay attention. But if you use a
relational database (the main case where it is interesting to just
have uft-8 strings), and set the encoding correctly everything works
just fine.

Philippe

>
> -- !
> Koji Yokokawa <koubo2006 at yengawa.jpn.org>
>     http://www.yengawa.com/
>     ^self new!
>
> _______________________________________________
> Seaside mailing list
> Seaside at lists.squeakfoundation.org
> http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside
>