[Date Prev][Date Next][Thread Prev][Thread Next][Author Index][Date Index][Thread Index]

Preliminary spec for Docs & Links layer !!!!! (yeah!)

The relevent directory is ~mark/xanadu/xpre-alpha/be
The relevent files are:

	Orgl*.*xx (now exposed)
	Hand*.*xx (Manipulators separated from Hands, Hands a subClass)
	Waldo*.*xx (Also a subClass of Manipulator. WaldoCategories are
	SubHand*.*xx (The foundation for manipulable multi-Orgls /
			works.  Support for abstraction promotion)
	Doc*.*xx (The Docs & Links layer itself (lower layers aren't
						 sealed off)))

Not only does the spec exist (really a bunch of protocol declarations)
but much of the support mechanism (stuff below Doc*.*xx) is working as
well.  More would be working if I wasn't having as much trouble with
dbx & dbxtool.  My apologies for how cryptic everything is: Roger &
Heh together should be able to decode it, & I'll do so as well when I
return from my cousin's wedding.  In fact, so much is working that
when I get back I'm going to shorten by *gleeful* estimate!
Speaking of symbolic debugging, Recently (thursday?), we found out
that G++ is safe after all.  The previous fear was due to a
misunderstanding about the word "run time libarary".  It turns out
that the code generated by G++ doesn't assume the existence of any
support routines which are copylefted.  They only use the support
library of GCC, which is public domain.  The misunderstanding was that
they (I'm not sure if the misunderstanding was at Hugh Daniel or at
Stallman) thought we were inquiring about the library of useful stuff
that comes with G++ (I/O, etc...).  That stuff is copylefted, but I
was totally uninterested in using that stuff anyway.

The reason I bring this up is that GDB (the Gnu symbolic debugger)
actually works with G++.  It actually lets the user use the
identifiers that appear in the source code and knows what they refer
to!  Amazing!  It is also supposed to work through emacs (meaning one
can visually debug from home).

G++ itself is supposed to correspond rather closely to C++ 2.0, and
they intend to track.  

Would anyone like to try compiling a recent pre-alpha with G++ and
reporting how it went?  Would anyone like to try using & learing their 
way around GDB?  I'd rather not be guinea pig on this; and Mr. Hill,
the other critical path, probably isn't the best choice either.  This
is a golden opportunity for non-critical path activity to shorten
critical path time.  Thanks.