Hi Timwi,

Le Thursday 01 July 2004 04:43, Timwi a écrit :

Yann Forget wrote:

...

I think we should have some filtrer for the mailing lists against spams and virus.

How come I never see spam or viruses here?

Is it because friendly gmane folk manually delete them from the server? If that is so, I can only recommend strongly everyone use gmane's NNTP interface too! :) If, however, gmane have automated spam filtering mechanisms installed, it means legitimate messages may sometimes not show up. Worrying...

Spams and virus are blocked because the sender's address is not subscribed to the list. There are dozens like this daily. And list admins have to clean to mess every now and then. It's becoming a unnecessary burden when virus and spams could just have been rejected because the address used is non valid.

Timwi

Yann

This one has been around for ages and is ''probably'' easy to fix. In Parser.php the function openList does this:

if ( "*" == $char ) { $result .= "<ul><li>"; } else if ( "#" == $char ) { $result .= "<ol><li>"; } else if ( ":" == $char ) { $result .= "<dl><dd>"; }

In most cases, this works fine. Consider the following list structure, however

* item *** 3rd-level item ** 2nd-level item * (more items)

Lists with this general structure are quite common (although editors usually go through some hoops to re-arrange them because of the problem I'll outline in a moment). The HTML that results, however, has issues:

<ul> <li> item <ul> <LI> <ul> <li> 3rd-level item</li> </ul> <li> 2nd-level item </ul> <li> (more items) </ul>

(Reformatted for clarity, and some leftover closing </li> tags omitted on the grounds that they don't do any real harm.)

Notice that spurious <LI> in line 4 (which I capitalised here to make it easier to see). It results in a very ugly display on the user's screen, looking rather like this:

* list item * * 3-deep list item * 2-deep item * (more items)

Notice the gap and the spurious list marker. The desired display is this:

* list item * 3-deep list item * 2-deep item * (more items)

How to fix it? I'm sure that someone with more skill than me could come up with a comprehensive and elegant fix, but here is a simple one on the brute-force level. Retain the basic:

if ( "*" == $char ) { $result .= "<ul><li>"; }

but reprocess it afterwards looking for the string "<li><ul>" and replacing the string with just plain "<ul>". If I have this straight in my mind, <li> should *never* directly precede <ul> in the output HTML. Similar logic applies to the other two list types.

Tannin (Tony Wilson)

On the contrary, Timwi,

Lists of the structure I cited are *common as mud*. Do you think I am a complete fool and just made something up? Shall I spell out a real-world example for you?

Consider a list of water catchment basins, which in turn lists the major and minor streams belonging to each one.

* Basin 1 ** major stream A ** Major stream B *** minor stream C *** minor stream D *Basin 2 *** minor stream E *** minor stream F * Basin 3 ** major stream G *** minor stream H

Note Basin 2 which has no major streams. What do you want the user to do? Go out and dig a new river so that you can have nice idealogically pure lists?

Got it? Or do you need another one?

* Cockatoos *** Cockatiel, Nymphicus hollandicus ** Black cockatoo group *** Red-tailed Black Cockatoo, Calyptorhynchus banksii *** Glossy Black Cockatoo, Calyptorhyncus lathami *** Yellow-tailed Black Cockatoo, Calyptorhyncus funereus ** White cockatoo group *** Gang-gang Cockatoo, Callocephalon fimbriatum *** Galah, Eolophus roseicapilla *** Sulphur-crested Cockatoo, Cacatua galerita

Examine the Cockateil (second line above). It doesn't belong to a sub-group

No, this is definitely not "desired", and I'm not even sure it's possible to do in all browsers; HTML is not a formatting mark-up language. It's a document structure mark-up language.

It definately *is* desired, chum.

This is *not* HTML, it is wiki markup It works just fine cross-browser (yes, including that perennial problem child Internet Explorer).

In any case, HTML does not exist to serve narrow ideaalogical purposes. It exists to serve *users*, in this case, users attempting to describe the real world. It is particularly important to remember the real purpose of wiki markup, because (unavoidably) the wiki markup language takes away the ability of the user to format pages using more flexible tools such as word processing software, HTML, or pen and paper.

Now, can we get back to the bug that I brought up and stop spouting nonsense?

Tannin (Tony Wilson)

On Fri, 02 Jul 2004 03:29:58 +0100, Timwi timwi@gmx.net wrote:

I'm pretty sure it is invalid HTML to put a UL directly inside a UL without an LI in between.

That is my understanding, too.

Cheers, Philip

If I remember correctly, my C++ parser ("waikiki" package in CVS) does this correctly. It remembers the list state of the last line parsed ("*" or "###" or ...), then 1. falls back with "</LI></UL>" (or the like) to the current level, if necessary 2. adds new "<UL><LI>" (or the like) to reach the current level, if necessary I also should handle mixed lists correctly ("**", then "###").

Those who dare read C++, have a look! :-)

Magnus

Brion Vibber wrote:

Philip Newton wrote:

...

On Fri, 02 Jul 2004 03:29:58 +0100, Timwi timwi@gmx.net wrote:

...

I'm pretty sure it is invalid HTML to put a UL directly inside a UL without an LI in between.

That is my understanding, too.

We went through this question a couple years ago and, yes, that's illegal. The correct way to nest is like this:

• Item

** Subitem

<ul> <li> Item <ul> <li>Subitem</li> </ul> </li> </ul>

Obviously this isn't very forgiving to skipping levels.

-- brion vibber (brion @ pobox.com)

---

---

Wikitech-l mailing list Wikitech-l@Wikipedia.org http://mail.wikipedia.org/mailman/listinfo/wikitech-l