first input character as it was attempting to use an uninitialized critical
section (mutex) while asserting the UART interrupt. As long as the UART
virtualization is enabled, this problem doesn't happen, even if don't actually
*use* the UART virtualization.

thing that has any telnet-fu.

SSH support seems more stable now too.

Currently there's not much difference between the two anyways.

Thanks Ragnarok!

data.

to be stuffed into the input buffer as NULLs.

The *ONLY* remaining broken bit is that it is ASSumed that output is in
cp437.  Since most of the stock fonts do NOT use cp437, this is obviously
broken.  However, until I start supporting unicode or multiple codepages
correctly, the world will just have to deal with this.

Input will only work if the character entered exists in the cp437 codepage.
If it doesn't, the input is ignored as if it never happened.  However, if
the character IS in the cp437 charset, it WILL work correctly now (cheers)

So, for example, on a Spanish keyboard, everything except the underlined
'a' and underlined 'o' (and the Euro if so marked at AltGr-E) will work
properly.

Thanks for everyones patience on this one... sorry you still can't have
a new build.  :-)

number (e.g. '1'), and the RX_BY_NUMBER mail server option was disabled, then
the mail would be rejected with "user unknown".
"name aliases" weren't effected by this.

Let's use /bin/sh as the default shell for *nix rather than for !(__unix__)

*cough* been there for almost a year now.

so you don't need to reload the content at any time.
This appears to have "fixed" the IE stuff, so add the <pre></pre> around
the body which IE needs since it apparently doesn't support
white-space: pre; in the CSS.

header and checking the If-Modified-Since header in code.
Speed up messages.xjs by only storing the header fileds we use rather than
the whole header.

parents.

Italics indicates that this message has unread replies
Bold incicates that this message itself is unread

sourcing in a <script> tag which doesn't generally work.

1)  Indicates if a message is unread according to the new scan pointers
2)  Indicates if any reply in the chain is unread
3)  Clean up the HTML a little
4)  Document the class hierarchy
5)  Add a login box (no, not a pop-up)
6)  Change the cursor to the "processing" cursor while loading stuff
7)  Make the groups a separate script so that they can be reloaded if the
    user changes
8)  Move the client-side javascript into a separate .xjs file
9)  Some style sheet cleanups
10) Ensure that some user is used... either "Guest" or the new login

Things I should be able to do fairly easily...

1) Have the new indication update dynamically so you can tell what you've
   already read
2) Update the new scan pointer when the user leaves the page
3) Add a reply button
4) Add moderation capabilities for the moderator
5) Add deletion support and a show delete messages option for operators
6) Also allow users to delete their own posts if allowed
7) Hook in the users email

What I can't/won't do

1) Work hard enough on the CSS to make it pretty
2) Make it look correct in IE (see #1)
3) Replace the current default system with it

#1 and #2 desparately need a web geek.

for it at the end of the line.

Prevent zero-length write()s in that case.

Use the infinately more correct toSource() method rather than the horrible
hand-rolled escape_quotes() function.

Also, add class info indicating if a message is new and if it contains
new replies.

Show total number of replies and newest reply date.