I told you it's in the NOTES file.
lotte% tail -1 /usr/fgb/src/abaco/NOTES
/sys/src/libhtml/lex.c:1258 if(c == ';' ) {
--- Begin Message ---
i did something wrong. you're right &pid is getting translated as π.
totally wrong,
since entities occur in html NOT urls. but anyway, what is your fix?
- erik
On Sat Jul 8 08:45:32 CDT 2006, [EMAIL PROTECTED] wrote:
> On Sat Jul 8 08:32:32 CDT 2006, [EMAIL PROTECTED] wrote:
> > > the offending url is:
> > >
> > > //adlog.com.com/adlog/i/r=7261&s=670913&x-fid=27&t=2006.07.08.12.32.41&o=1:&h=cn&p=2&b=5&l=en_US&site=3&pt=2001&nd=1πd=&cid=0&pp=100&rqid=00c17-ad-e444AE8131A6EF99/http://i.i.com.com/cnwk.1d/Ads/common/dotclear.gif
> > >
> >
> > did you apply the change?
>
> what change?
>
> > according to russ href's cannot contain '&' they should have "&"
> > instead,
> > but this is not the rule out there. libhtml code sees '&' and doesn't wait
> > for the ';'
> > so "&sp" becomes ' ', with my change "&sp" is still "&sp", but "&sp;" gets
> > translated
> > into ' '.
> >
>
> hrefs *MAY* contain &. & is the argument sepertaor for cgi scripts. if you
> have
> an "&" in an argument, you need to url-escape it -- that is a much different
> escaping
> mechanism than for html. for example "&" would be encoded as "%26".
>
> it is curious that they are not using a ? to seperate the base url from the
> arguments.
>
> i used curl on my linux machine to inspect the text sent from the news.com.com
> server. the url really does have a π in it. try it out. also, html
> entities like &sp; or π should never be interpreted within urls.
>
> - erik
--- End Message ---
