a new DownloadService to detach TiddlyWiki from the web

[BidiX]

As discussed in this thread: http://groups.google.com/group/TiddlyWikiDev/browse_thread/thread/9c0a4a9e989fce4f it could be a valuable improvement if the ''engine'' (aka the ''core'') could be put in a separate file. Especially for those of us that access their TiddlyWiki by majority on the web (TiddlySpot users for example).

BramChen maintains a distribution on this base (empty.html and TiddlyWiki.js  in PrinceTiddlyWiki distribution). And everything go right when accessed from the web and saved to web using UploadPlugin. And we could save 220Ko on each download and upload (the size of the actual core).

But the challenge is to maintain the single file paradigm of TiddlyWiki and be able to detach it from the web. To do that :

• DownloadService is a script (download.cgi) that includes the javascript files that are referenced in the html file.
• UploadService (store.cgi) has been modified to do the reverse (ie replace a previously inserted content by its corresponding filename).
  

See http://TiddlyLab.bidix.info/ for this experiment.

Advices and suggestions are welcome.

-- BidiX
http://TiddlyWiki.bidix.info

[Clint Checketts]

I'm looking at it right now, but if this does what I think, then life just got a whole lot better.

Thanks BidiX!

-Clint

[Ken Girard]

All I can say is "Wow".
Is there anything that you guys can't get TW to do?
And this took you just a couple of hours...
OK, now grant my next wish. I want a pony!
Now where am I going to put a pony...

Ken Girard

[Ken Girard]

OK, I have to ask. What kind of improvement in load time are you
getting with this setup?
And how long before everything tagged systemConfig gets saved as a
seperate js file?

Ken Girard

[Hans]

... and maybe systemStructure (or similar) that could take care of
themes, menus etc.

Just some toughts.

[BidiX]

A small revision of DownloadService: download.cgi now accept to include a remote script when you detach your file from the web server ( ie like <script src=" http://tiddlyhost.bidix.info/lib/TiddlyWiki_2.1.0.b4.js" ...).

Therefore it is possible to centralize the 'core' script for a whole domain. Only one instance for all subdomains of a TiddlyWikiHosting domain (like TiddlySpot.com) and for the browser only one version in the cache.

A step further, why not a single instance of core used by all TiddlyWikis located at TiddlyWiki.org ... Thoughts ?
An other step more ...  to include directly plugins from author's sites ... or a certified repository ... Thoughts ?

New revision (v1.1.1) at http://TiddlyWiki.bidix.info/#download.cgi

-- BidiX
http://TiddlyWiki.bidix.info

[Greg Wolff]

BidiX,
Good idea. A certified repository is exactly the kind of thing we're
working to build and support on TiddlyWiki.org.

What kind of interface would you want to access plugins through such a
repository? Would you pull them in from individual .js files each on
their own URL? Or would you want a script interface for retrieving a
bundle at a time?

Thanks for download.cgi and helping support the unplugged spirit that
distinguishes TiddlyWiki.
-Greg

[BidiX]

Hi Greg,

On 8/19/06, Greg Wolff <greg....@rdvp.org> wrote:

What kind of interface would you want to access plugins through such a
repository?  Would you pull them in from individual .js files each on
their own URL? Or would you want a script interface for retrieving a
bundle at a time?

 For now, I've no solid proposition to make. I've experimented to import tiddlers by RSSFeed, but it adds a new level of format and protocol. Since ImportTiddlers seems to be well installed in TiddlyWiki, accessing an html storeArea, interactively or in asynchronous mode, is certainly a good way. I'll certainly experiment in this direction soon. 

Thanks for download.cgi and helping support the unplugged spirit that
distinguishes TiddlyWiki.

 I support unplugged and single file spirit of TiddlyWiki too ;)

-Greg

-- BidiX
http://TiddlyWiki.bidix.info