On 8 June 2014 12:18:49 BST, Bob Latham <b...@mightyoak.org.uk> wrote: >In article ><out-539399f2.md-1.4.17.chris.yo...@unsatisfactorysoftware.co.uk>, > Chris Young <chris.yo...@unsatisfactorysoftware.co.uk> wrote: >> On Sat, 07 Jun 2014 19:58:41 +0100, Richard Torrens (lists) wrote: > >> > After a bit more experimenting, it is caused by the line: >> > >> > <P><a href="http://<!--#echo var="HTTP_HOST" --><!--#echo >> > var="DOCUMENT_URI" -->">Top of page</a> > >> Ah, right. I know exactly what the problem is and what causes it. > >> NetSurf now rejects invalid URLs. By "invalid" I mean, anything >> containing characters that are not permitted either in old-style >ASCII >> names, or in IDNA, plus some other validity checking mostly related >to >> IDNs. > >> Previously NetSurf would percent-encode the host and return that, so >> you'd end up with a garbage URL. Now it rejects the URL outright. >> However, it appears the layouting can't cope with that. I did some >> extensive searching when I made this change initially and was unable >> to track down why this was causing the layouting to fail - ideally, a >> broken URL shouldn't cause termination of the layouting process, it >> should only make the problem link go nowhere. > >> I considered this low risk at the time as it is incredibly rare to >> encounter a link to a URL which doesn't conform to the rules. > >> However, obviously it does need to be fixed, so if you can raise it >on >> the bugtracker that'd be grand. > >Good luck with that. > >I'm plagued with this BocConvert as well as told in my thread "Username >and Password problem". > >Daniel showed some interest and asked for information and I've tried >everything to get it to him and after (in total) a couple of hours >trying >I've given up. > >The log from my Netsurf seems to say that it bombs out BoxConvert >because >it encounters the character 5f presumably hex or &5F. I have searched >through the relevant html on my intranet and no 5F anywhere and >certainly >not in a page url. > >Mine is triggered not by the web page as such but by the security >username >and password on items inside one folder on my Cherokee driven Intranet. > > >This started to happen for me with build 1950 and has persisted ever >since >but builds up to and including 1949 did not have the problem. I would >not >be surprised if the problem in this thread started with also 1950. > >As regards reporting the problem, I found this an obstacle course and >no >doubt. > >Firstly, I was asked for a log file from netsurf which after some >reading >and testing I managed to generate. I possibly wrongly emailed it to the >mailing list. I was told it was being help for the next moderator as it >was a little too big. > >So I emailed Daniel to his email address with the file, no reply. > >He also asked me for something in cgi. Err what? > >And the config file from Cherokee for which I've asked what's the file >name and where it may be in the structure but had no reply. > >The I thought maybe I should look into this bug tracker whatsit, that >when >the problems started. > >Registration kept bombing out or doing nothing and I had to use a PC to >get this done. I then logged in using the ARMiniX and Netsurf filled in >a >report and then found in Netsurf I could not attach a log file. > >I logged in on the pc, filled min the forms all again, attached the >file, >hit send and it bombed out again. > >At this point I walked away pretty wound up if I'm honest.
Apologies for the excessive quoting. Can you send the log to me? Chris
