Curl Developer Community : All Content - All Communities

Search function

mdsuka@gmail.com — Thu, 28 Jan 2010 05:57:56 GMT

Search problem is from few month ago and still not working.

ok good luck for rebuilt.

About reloding a package

tdeng@curlap.com — Tue, 19 Jan 2010 09:38:15 GMT

I haven't yet fully understood what's your real problem is, becuase there are many factors that could affect your result.

The special way I wrote above is to ensure that the file will be re-downloaded from web server, instead of the one that has once been cached in browser cache on Windows, and this is usually not needed if you configured your settings related with resynchronization correctly for your Curl applet, including settings in your source codes and settings in Curl Control Panel on your client RTE.

Did you try with settings like the following in your applet ?

{applet
resync-as-of = {utc-date-time "2003-10-11"}
}

Timer and memory

alchimiste@paran.com — Tue, 19 Jan 2010 04:24:44 GMT

Thanks a lot.

My customers often ask me about RTE memory use, and somtimes they monitor Window's task manager.... that drives me mad.

Is there a need for a fire wall rule or is this somethig else?

curler@westboro.org — Mon, 18 Jan 2010 23:03:25 GMT

Thank you both. You have answered my question. I will allow the communications to happen - it just caught me offguard. It would of been nicer if teh reverse lookup was more reasonable - frankly it looks like an automatically generated dial up address.

Bob

text formats

admin@communityxperts.com — Sat, 16 Jan 2010 23:18:40 GMT

Anyone interested in related text issues might like to look at this Pango and related link which I have found useful on the Gnu/Linux side.

Is their any APIs like LineNumberReader in Java ?

tdeng@curlap.com — Wed, 06 Jan 2010 00:29:39 GMT

Thanks Chris and Robert. I think Chris's ZUZU could help me get what I want.

Linked Lists

cbarber@curl.com — Mon, 04 Jan 2010 21:55:54 GMT

Dequeue-of probably has an overly minimal interface, but the ZUZU.LIB.CONTAINERS package has an extended subclass that provides methods for insertion and removal of elements. Of course, because Dequeue-of is implemented as an array, insertion and removal anywhere other than the head or tail costs O(n).

It really doesn't make sense to try to construct a linked list node class in a statically typed inheritance based language such as Curl because such a class would be overly restrictive. The prev/next pointers would have to have the static type of the base class, and thus all subclass methods that operate on them would end up having to do explicit casts to get the subclass type. It ends up being less work to roll your own. One thing you can do is to write a macro that will generate a linked list API in a given class, e.g.:

{define-class MyNode

field x:int, y:int, z:int

{generate-linked-list-api}

}

This can generate the prev/next fields along with any helper methods you feel would be useful. The question is what exactly should that API look like.

The ZUZU.LIB.SYNTAX package does have a couple of macros that can be used for iterating over a linked list: for-list, for-prev, and for-next. E.g.

{define-class MyNode

...

field next-node:#{this-class}

...

{method {do-something}:void...}

...

}

{for-list next-node n in my-list do

{n.do-something}

}

Trees and Multicore

admin@communityxperts.com — Sat, 02 Jan 2010 15:28:24 GMT

At the moment, Curl 7.0 offers the LazyTreeNode class to provide a performance alternative when it is required by the nature or use of a tree (those nodes may be best suited to read-only TreeModel's.)

Multicore CPU's might offer another performance alternative for populating trees with an async refresh/fully-populate option.

Curl Browser

admin@communityxperts.com — Fri, 18 Dec 2009 16:44:18 GMT

The EU settlement with MS should give a boost to alternative browsers such as Opera and, hopefully, to site-specific and task-specific browsers.

Now that DCurl deploys have placed Curl on the desktop, it might be time to consider more than just wrapping a library such as WebKit to achieve a Curl web browser (we currently have the embedded browser widget but with Mac and linux issues/limitations.)

One consideration could be this: that a Curl browser's priority should be that of consuming Curl - and not necessarily via HTTP.

The advantage of wrapping a library is that it makes it possible to wrap both an HTML library and an ECMAScript library such as SEE (Simple ECMAScript Engine.)

An alternative approach might be for the Curl Browser to consume JavaScript objects as Curl closures - and where possible, to replace embedded JS functions with Curl procedures. Page input restricted to ECMAScript sources using jQuery or DOJO (or some eventual standard) as libraries might be doable in a first release.

Furl is not available as a name (and we don't want to furl our curl sail.)

Kihara and Kiharin look to be available (see Finnish "kiharaa" ) or Japanese "Karu"

Server-side Curl

admin@communityxperts.com — Fri, 18 Dec 2009 03:44:18 GMT

One prospect for re-thinking server-side versus client-side for high-performance and realtime applications would be to embrace a language-neutral 2-way network service framework such as Cisco Etch (now in the Apache.org incubator.)

Etch is currently C# and Java and is expected to soon embrace C, JavaScript and Python. Before Scala and Clojure get to Etch, Curl could be there. Before HTML5 adoption and before ECMAScript Harmony. Not that Etch is the only strategy.