🚀 go-pugleaf

andrew abernathy writes
| 	I have a (hopefully) quick question about matrices, cells and  
| tags.  I use browsers in several places to allow the user to select some  
| entity.  When I load the matrices, I set the cells tags to unique values  
| so that I can easily identify a selected cell.  Well, I can get the cell  
| that is being selected, but it's tag always returns as -1.  I'm having to  
| identify cell by their stringValues, which is very inconvenient.  Does  
| anyone have any ideas why I might be getting the behavious I am?
| -- 
| andrew_abernathy@wire.gwinnett.com (NeXTmail gleefully accepted)

I am very familiar with this NXBrowser problem and I know exactly what the  
cause is (although I don't know why NeXT designed it this way).  If you look  
closely at the NeXTstep References for the AppKit classes, you'll see that  
NXBrowserCell is a subclass of Cell.  If you read the -tag and -setTag: methods  
for the Cell class, you'll note that they are not implemented for some reason.   
Instead, they are designed to be overridden and implemented in subclasses of  
Cell.  If NeXT had written NXBrowserCell as a subclass of ActionCell instead of  
Cell, then your code would work as you wrote it.

The solution, which I have working under System 2.2a, is to subclass both  
NXBrowserCell and NXBrowser.

You can't make a category of NXBrowserCell since you need to add an instance  
variable to hold the tag.  So, in your subclass of NXBrowserCell, just add an  
instance variable named "tag" and write the (obvious) code for the -tag and  
-setTag: methods.  You should also override the -initTextCell: method and  
initialize this instance variable to something predictable like -1.  I skipped  
the implementation of -read: and -write: since I never archive my browsers  
(other programmers may need to fill in these last two methods as well).

You also need to write a subclass of NXBrowser because I have found no other  
reliable method of setting the cell class and cell prototype (i.e. one that  
doesn't eventually cause the program to crash).  You must call -setCellClass  
with the name of your NXBrowserCell subclass within the +initialize method for  
your NXBrowser subclass.

#mount _soap_box
This is a prime example of how NeXTanswers is nearly useless.  appkit.827 says  
to -setCellClass inside -initFrame:, appkit.816 says to -setCellClass in  
-appDidInit:, appkit.761 says that you must -setCellClass in +initialize  
because -init is too late, implying that -appDidInit would be far too late.  It  
took a great deal of experimentation and searching to figure out what worked.   
Of course the simpler approaches didn't work and they were what I tried first.
#dismount _soap_box

Within my main program, I call -setCellPrototype before using the browser.   
This can probably be done in -appDidInit, but since I have several .nib files I  
use the -setBrowserOutlet initialization method for the nib owner to accomplish  
this (where browserOutlet would be the name of my instance variable).  My only  
remaining problem is determining whether I should worry about -free'ing the  
cell prototype when I free the nib instance.  I could assume that NXBrowser  
will free its cell prototype within its own -free method since the program  
crashes when I -free the cell prototype myself.  If anyone reads this far and  
can offer a hint, I would appreciate it.  MallocDebug finds so many other  
memory leaks that I can't tell whether NXBrowser is freeing the prototype cell.

Sorry I didn't post source code, but this stuff is not packaged as an example.   
It is hidden deep within a program that I am currently writing.
-- 
Brian Willoughby	Software Design Engineer, BSEE NCSU
BrianW@SoundS.WA.com	Sound Signal Processing Software and Consulting
NeXTmail welcome	- NO EMAIL SOLICITATION without prior permission

My architecture is the following:

Client app A
Server app B

Server B is running, client A gets a connection to B via 
distributed objects. B has registered a named object, and
runs the connection via runFromAppkit.

Client A calls [B addClient:self] which succeeds, and A is
added to a List object of clients of B.

Now B starts up a DPSTimedEntry, which generates events every
30 seconds that are supposed to be passed on to every object
in B's client list. However, these messages never return,
and the client app doesn't recieve them. Every now and
then an "Unrecognized error" pops up in the console.

When the client app terminates, about 1/3 of the messages
that were being sent (about 3 minutes worth) suddenly
are recieved (I suppose it occurs when the connection
is freed by the Application Object).

In any event, everything seems to run fine if the DPSTimedEntry
isn't generating the events (that is, if the methods are called
from appDidInit: or something...) 


So, is this me forgetting to do something, or what? Any help
is appreciated!

- darcy