I agree with everything Stephen and Erwan have said.  We should feel free to make syntax changes pre-1.0 to ensure we have the cleanest approach going forward.  I've found during a migration from SnipSnap that it's not too difficult to whip up some Groovy that filters old to new.

I would like, though, to see the API expose a DOM of sorts.  This would make migrating between syntaxes quite robust (use the "old" renderer to produce DOM, write out with "new" renderer) and the effort of opening up the rendering process would enable sectional editing and other fragment-context functionality later on.
- - -
Hans Gerwitz


On Oct 28, 2005, at 11:59 AM, Erwan Arzur wrote:

Hey Stephen

we recently had a discussion with Ludovic about choosing a rendering engine and not mix both velocity and groovy.

About your UI proposal, i think it needs some more thinking because we might want to plug other engines in addition to groovy and velocity, but i find it better and easier to implement than my own idea which was to implement a kind of preprocessor which would allow the user to select the rendering engine to use for her document.

agree with your proposal about {pre} and {code} radeox tags, with the provision that it might prove very complex to hide selected content from the velocity/groovy engines.

Agree about error messages, they need a lot of improvements

And you're right about all those not fitting into a single JIRA issue :D

1. choosing a rendering engine. I'd like to add that mixing the two in a document is a very bad design decision (but facts could prove me wrong), just for the possible namespace collisions that could happen (your $varname example)
2. {pre} syntax not coherent with other tag's
3. {pre} {code} specs should be amended
4. forcing enclose scripts into radeox tags

All this make significant syntax changes to make me wonder about existing documents ... how to migrate them or make sure the old syntax is still ok for them ?

Erwan

On 10/28/05, Stephen Schaub <stephen_schaub_88@hotmail.com> wrote:
When XWiki renders a document, the interactions between the Groovy,
Velocity, and Wiki (Radeox?) syntax processing can lead to unexpected and
undesirable behavior. There have been posts on this list in the past where
this issue has been discussed. However, I'm not aware that any consensus has
emerged as to how to deal with these problems.

To summarize, here are a few of the issues that I'm aware of:

* Getting XWiki to render a bit of text without doing any syntax processing
on it can be difficult. For example, the {pre} ... {/pre} markup turns off
XWiki syntax processing, but not Groovy processing.

* The # symbol is used both to do numbered lists and is also used for Groovy
processing. If a user wants to create a numbered list and forgets to put a
space after the #, he can run into trouble. The following example will cause
a stack dump:

  #bring coffee
  #include doughnuts

* Code samples on a Wiki page sometimes don't render correctly, due to
conflicts with Velocity syntax processing. The following C/C++ code sample
results in a runtime stack dump:

{code}
#include <stdio.h >
...
{code}

* Groovy and Velocity both use $varname syntax. I think I've read some posts
in the past where this was an issue -- although I can't come up with any
examples at the moment.

* A usability issue: Most XWiki tags ({table}, {code}, etc.) come in pairs.
The closing tag has no slash. For example:  {table} ... {table}. However,
the {pre} tag does have a closing slash: {pre} ... {/pre}. This
inconsistency is annoying and causes difficulties for new users.

* A significant usability issue for Windows users: Pathnames contain
backslashes (\). XWiki uses the backslash as an escape character. So
attempting to put a UNC path like  \\myserver\myshare\mydoc.txt in an XWiki
document triggers an XWiki runtime exception.  Boy, is that frustrating for
new XWiki users. The solution is to wrap the path in {pre} .. {/pre}, but
the error message gives the user no clue what the problem was, much less how
to correct it.

Here are some proposals for discussion:

1. {pre} ... {/pre} should turn off ALL syntax processing of its contents.
2. {code} ... {code} should turn off ALL syntax processing, except for
syntax coloring.
3. {pre} should allow {pre} as a closing tag.
4. Consider requiring users to explicitly enable Groovy/Velocity processing
for selected Wiki documents as needed. In the page editor, provide two
checkboxes: "Perform Groovy Processing" and "Perform Velocity Processing".
The user could separately enable either Groovy or Velocity processing, or
both. People who enable them would presumably be in a better position to
deal with the kinds of syntax conflicts that would occur.
5. Provide more user-friendly error messages when a Groovy or Velocity
processing exception occurs. Show the Wiki source line that caused the
problem.

I haven't opened a JIRA issue for this yet, because I'm not sure how best to
word it. But I think this is a high-priority issue that should be carefully
addressed before the 1.0 release. This may be an issue that warrants a page
on the xwiki.org developer site. I'll be glad to start one if the developers
desire it -- perhaps here:
http://www.xwiki.org/xwiki/bin/view/Dev/Discussions

Stephen






--
You receive this message as a subscriber of the xwiki-dev@objectweb.org mailing list.
To unsubscribe: mailto:xwiki-dev-unsubscribe@objectweb.org
For general help: mailto: sympa@objectweb.org?subject=help
ObjectWeb mailing lists service home page: http://www.objectweb.org/wws




--
You receive this message as a subscriber of the xwiki-dev@objectweb.org mailing list.
To unsubscribe: mailto:xwiki-dev-unsubscribe@objectweb.org
For general help: mailto:sympa@objectweb.org?subject=help
ObjectWeb mailing lists service home page: http://www.objectweb.org/wws