Lisp HUG Maillist Archive

processes and pointers

I'm investigating a "non-garbage garbage" problem, a case where the GC
didn't want to remove some hundred megabytes of data which, I thought;
wasn't referenced at all.

The data was collected by the GC as soon as I removed references to a
certain clos object. This object didn't - I think (I may have missed
something, I was in a real hurry to fix a server that was bogged down
by its own weight) - have any references to the large data structures
that weren't collected.

But it _did_ have a reference to a dead process, so here's my
question: Can dead process objects still contain internal references
(that aren't visible to the inspector...) to, e.g., dynamic bindings
of variables, and thus bar data from being GC'ed?

(Using LWL 4.2.7)
-- 
  (espen)


Global preferences etc.

In LWW 4.3.6, I want to save more state between
sessions than what is presently saved of the global
preferences.  For example, if I have a listener
maximized in the mdi window, and I quit and restart,
the listener will no longer be maximized.

If I wanted to save that particular detail to a file
and restore it on restart by loading that file from my
..lispworks, how would I do it?  Would I find it in a
specific global variable?  Or would I have to get it
from the Microsoft Window data via FLI?

Another such detail is the space occupied by the
buttons below the menu.  When deleting all those
buttons, the space is reduced to a thin line, which is
what it should be in that case.  But after restarting,
the space is restored to its original size, even
though the buttons are still removed.

I'm also wondering if I could write a function,
find-related-symbols, which would take a symbol as its
argument and return a list of related symbols.  It
would work as follows:  First, it would make a list of
all functions that referenced the given symbol.  I
don't know how to do that, but I assume it's possible.
 Second, it would make a list of all symbols
referenced by all of those functions.  Third, it would
count how many references existed in the image to each
symbol in the list, and omit those with too many
references, on the theory that they would probably not
be very closely related to the symbol in question.

Would such a function be trivial, or very difficult,
or what?  Any hints how it would do the cross
referencing?  Or does such a function or something
similar already exist?  As an example of what I would
do with such a function, suppose I wanted to find the
global preferences, to figure out how to control them
from my .lispworks or whatever.  I might give this
function a symbol such as :MDI, which is probably
related to the global preferences, and that might be
the fastest and easiest way to find them.


__________________________________
Do you Yahoo!?
Protect your identity with Yahoo! Mail AddressGuard
http://antispam.yahoo.com/whatsnewfree


Re: Global preferences etc.

Unable to parse email body. Email id is 1512

Re: Global preferences etc.

Unable to parse email body. Email id is 1526

Meta Key Under X11 in LW 4.3.6

In LW 4.3.6 under Mac OS X 10.3, openmotif 2.2, and Apple's X11 v1.0, I'm trying to figure out how to get option to be the meta-key in preparation for doing some CLIM hacking.

Right now, I have to type either escape or command.

Should I be looking at lispworks parameters or setting the xmodmap?

Any clues?


Re: Meta Key Under X11 in LW 4.3.6

Thanks Dave.  This works fine.

Is there a recommended way to store these key mods in some init file
so they don't have top be typed at every start up? For example,
a LispWorks specific X init?

At 12:43 PM +0000 11/13/03, davef@xanalys.com wrote:
>   In LW 4.3.6 under Mac OS X 10.3, openmotif 2.2, and Apple's X11 v1.0, I'm trying to figure out how to get option to be the meta-key in preparation for doing some CLIM hacking.
>
>   Right now, I have to type either escape or command.
>
>   Should I be looking at lispworks parameters or setting the xmodmap?
>
>xmodmap
>
>   Any clues?
>
>I mapped the Option key's keycode to Meta_L by
>
> xmodmap -e 'keycode 66 = Meta_L'
>
>and then did this to add Meta_L to mod1:
>
> xmodmap -e 'add mod1 = Meta_L'
>
>LispWorks started then responds to Option as the meta key.
>
>
>FYI, if I have Meta_L on mod2 and switch off "the Enable key
>equivalents under X11' preference in X11.app (v0.2) then LispWorks
>responds to Command as the meta-key.
>
>--
>Dave Fox			
>
>Xanalys                 http://www.lispworks.com
>Compass House
>Vision Park, Chivers Way
>Histon
>Cambridge, CB4 9AD
>England


Re: Meta Key Under X11 in LW 4.3.6

Unable to parse email body. Email id is 1534

LW Documentation Under X11 in LW 4.3.6

The next question is how should the browser variable be set in order to
show LispWorks documentation under X11? 

If we have an X11 browser, then everything works for a remote user as well. 

Which X11 browser is recommended for LW documentation purposes?


Re: LW Documentation Under X11 in LW 4.3.6

Unable to parse email body. Email id is 1538

Re: LW Documentation Under X11 in LW 4.3.6

At 10:48 AM +0000 11/14/03, davef@xanalys.com wrote:
>   The next question is how should the browser variable be set in order to
>   show LispWorks documentation under X11?
>
>   If we have an X11 browser, then everything works for a remote user as well.
>
>   Which X11 browser is recommended for LW documentation purposes?
>
>Netscape.

But does it run under X Windows?

How should it be invoked to run under X Windows?


Re: processes and pointers

Unable to parse email body. Email id is 1508

Re: processes and pointers

davef@xanalys.com writes:

> No, dead processes should not have stacks, so the dynamic bindings
> will be gone. Perhaps the data was referenced via the
> process-arguments or process-plist, which might contain any CAPI
> interfaces that were created by the process. I don't suppose you got
> to take a close look at the dead process object?

Unfortunately, I was too much in stress to find out what was going on,
and all attempts to reconstruct the situation in the test environment
have failed - so far :-(
-- 
  (espen)


Updated at: 2020-12-10 08:59 UTC