Re: NSBrowser and lots of data
Re: NSBrowser and lots of data
- Subject: Re: NSBrowser and lots of data
- From: Tim Davis <email@hidden>
- Date: Sun, 29 Jul 2007 20:14:54 -0400
Breaking into simpler structure? Like...breaking each large Xml
fragment into it's own NSXmlDocument. Wouldn't that be the same
memory footprint? Is XPath slower than just navigating the
NSXmlElements using normal methods?
On Jul 29, 2007, at 8:07 PM, John Stiles wrote:
Have you considered preprocessing the XML into a simpler structure
in memory instead of using XPaths to do dynamic queries as the
program runs? If your concern is efficiency, it seems like this
would be the biggest win you could make.
On Jul 29, 2007, at 5:03 PM, Tim Davis wrote:
Hello all (again),
So I'm working on my first ObjC/Cocoa application after giving
up .NET and I was wanting to see what advice, if any, you guys
could give me. I've got my program working currently but it
seems...hrmm..slow and memory intensive.
Purpose of the software:
This is a type reflector tool for OS X, mainly to be used by .NET/
Mono programmers. It shows reflected types/classes/methods
etc...from .NET assemblies. I use the doc output from monodocer
to populate a directory of files, within these files is the xml
that describes the class and all of it's available public and
protected methods. I know...you guys aren't interested in .NET
but I just wanted to explain clearly what the goal is.
What I'm doing now:
The outputted directory structure generated by the monodocs tool
isn't that easy to navigate. In .NET a sample namespace is
System, another namespace is System.Xml, and it can also have more
(ie. System.Xml.Schema).... The optimal favored directory
structure would be /System/Xml/Schema for example. But it just
creates a folder called "System.Xml.Schema".
Currently what I do, once the docs are generated, is I parse
through them and grab the relevant info. (ie, Namespace, Scope,
Methods/Properties) and write all this data into one large XML file.
Here's an example for the namespace "ShadowMud.Systems.CommandSystem"
<Root>
<Namespaces>
<ShadowMud>
<Systems>
<CommandSystem>
<CommandController>
<Member Type="Constructor"
Scope="public">.ctor</Member>
<Member Type="Method" Scope="public">Add</
Member>
<Member Type="Method"
Scope="public">Count</Member>
<Member Type="Property"
Scope="public">CurrentCommand</Member>
<Member Type="Event" Scope="public">Error</
Member>
<Member Type="Property"
Scope="public">Processing</Member>
<Member Type="Method"
Scope="public">Start</Member>
<Member Type="Method" Scope="public">Stop</
Member>
</CommandController>
<CommandObject>
<Member Type="Constructor"
Scope="public">.ctor</Member>
<Member Type="Constructor"
Scope="public">.ctor</Member>
<Member Type="Constructor"
Scope="public">.ctor</Member>
<Member Type="Property"
Scope="public">Arguments</Member>
<Member Type="Property"
Scope="public">ClientID</Member>
<Member Type="Property"
Scope="public">Command</Member>
<Member Type="Property"
Scope="public">Prompt</Member>
</CommandObject>
</CommandSystem>
</Systems>
</ShadowMud>
</Namespaces>
</Root>
There's alot more than that, I don't want to liter your email
boxes with needless junk. After I parse all that into the file,
the file is loaded into an NSXmlDocument.
I retain that XmlDocument for use as the user navigates the
NSBrowser. The NSBrowser is setup so each column is the Namespace
contents, or class contents of the previous column's selected
item. I do this by simply grabbing the pathToColumn, and
coincidently that is a valid XPath I can use to gather the nodes
that are needed for the column being currently displayed.
In the following xml example, Click on ShadowMud, new column
contains 1 item (CommandSystem), click on CommandSystem new column
contains 2 items (CommandController, CommandObject), click on
CommandController and new column contains the <Member>'s for that
type. Pretty simple right?
In my NSBrowser delegate I just use the XPath to return all
matching nodes from the xml tree and return objectAtIndex
depending on what row it's asking for...I'm thinking that this is
the slow down, if a column has 100 rows, it's querying this a
hundred times...
Here's an example screenshot of the current work finished...(PS In
the screenshot I'm in System.Xml, XmlDocument is a type (class) so
the next column shows it's members.
http://www.binarymethod.com/files/images/node45.png
What am I asking???:
Clicking on things and the new column opening up performs very
slowly. Also my simple app is hogging 53 Mbs of memory. I was
wondering if any of you had any ideas, or advice on how to keep it
speedy???
Here's some ideas I've had..
1) Rearrange directory structure so it follows namespace paths
(ie. /System/Xml/Schema) and use a NSFileManager to tell the
NSBrowser what goes in what row in what column.
2) I've read a little on binding xml trees, but I have no clue
what binding is or how to use it and the docs I find aren't very
informative for a newb like myself.
....Well, if you read all of that then I thank you very much and
please shout out any ideas you might have to make my memory
footprint lower and speed it up...
Thanks
_______________________________________________
Cocoa-dev mailing list (email@hidden)
Please do not post admin requests or moderator comments to the list.
Contact the moderators at cocoa-dev-admins(at)lists.apple.com
Help/Unsubscribe/Update your Subscription:
40blizzard.com
This email sent to email@hidden
_______________________________________________
Cocoa-dev mailing list (email@hidden)
Please do not post admin requests or moderator comments to the list.
Contact the moderators at cocoa-dev-admins(at)lists.apple.com
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden