Xinha Discussion Forum

These problems should be fixed now.  Sorry for the delay.

Very kind of you, xinha.org fits

We don't need that many resources actually, it's just the fact that we need to install subversion (and have it thusly open on ports) and python, and ... otherwise it would be fine even on a shared host (and I'd host it myself in that case), we don't have huge traffic, or CPU use or memory use.  Yet.

Should apply from day 1, it was the way I wrote it initially

Can you submit that as a ticket please ("New Ticket" at the top of the page).  Thanks.

Unfortunatly because it keeps getting vandalised we can't have the wiki open to all.

I could maybe, possibly, perhaps open it to all members who are subscribed to the forum, by way of making the trac authentication use punbb's member information (actually, get apache to use it, trac leaves auth up to apache).

I'll look into it.

Hmm.  updateToolbar() maybe, but it should happen automatically

xinha_editors['page' + n + 'ID'].insertHTML()  - after you have done the make_editors call, xinha_editors actually contains the editors themselves.

I have plans in my head, and there is some code in the source.

If you have the time, js knowledge, some php/perl knowledge and a bit of inspiration, feel free to take it from where I left it, I don't know when I'll get a chance to get back to that area.

See htmlarea.js, search for "loadlang", and have a read of the _lc method below it.  Basically loadlang should return, for a given "context" an object containing the translations for a string

{ "Hello" : "Bonjour" }

Then the idea is that where translations are required you do

HTMLArea._lc('Hello', 'HelloWorld')

and it returns either a translated string for "Hello" in the "HelloWorld" context.

The context defaults to HTMLArea, and is intended to basically be either the default or the name of a plugin.

The central point of the plan is that I want to be able to use loadlang both with static javascript translation files like we have currently, but also be able to "target" it to a php backend using GNU gettext, or a CF backend using some special database, or a ..... you get the idea.

The bit that needs some thought now though is the *default* behaviour of loadlang, specifically, given a context, how should it figure out what static javascript file it's supposed to load (using _geturlcontent()).

You should also have a look at the Linker plugin, and particularly inline_dialog.js (in the root), this already uses the _lc function to get translations, although it will only ever return the untranslated strings currently.

In what way is EP broken?  It seems to work ok for me (after the patch a while back to fix the double-paragraph problem).

Yea, I was pretty sure that was how it worked.  So effectily, in word we get <p>Foo</p><p>|</p> when hitting enter and <p>Foo<br/>|</p> when hiting shift-enter, but it doesn't display breaks between the paragraphs.  I wonder what the internal representation of enter-enter is, <p>Foo</p><p></p><p>|</p> ?  It think it might be, because in a list if you do enter-enter you'll get two list items (ie two paras, an empty one and the current one).

Well.  I don't know.  Which do you want to implement?

Dong the <p></p> => <br /> (or <p> </p> - a whitespace paragraph rather than a null one) would be easiest, a simple dom search when you hit enter.  Doing <br /><br /> => <p> would be hard to get right I think.  Adding spans all over the place is just freakin ugly IMHO, but again, I just wouldn't use the plugin so doesn't matter to me.  Adding classes or styles is similarly ugly.

Classes however may be the way..

Hit enter once and we get
<p class="noBlankLine">Foo</p>
<p>|</p>

hit again and it becomes
<p>Foo</p>
<p>|</p>

so hitting enter in the paragraph would have to detect if the last keypress was also enter,
yes: grab the previous sibling paragraph, remove the class, cancel the enter
no:   set the class on the current paragraph, allow the enter to continue (return true)

I think this would be relatively easy.

Could you not do something with $PHP_SELF to work out where it's installed?