Experimental Physics and
Industrial Control System

[email protected] (Rozelle Wright) · Mon, 5 Jan 1998 08:31:21 -0700

> From [email protected] Fri Jan  2 12:14 MST 1998
> Date: Fri, 2 Jan 1998 09:14:10 -1000
> To: [email protected]
> Subject: e2db changes
> X-Sun-Charset: US-ASCII
> Content-Type: text
> Content-Length: 2901
> X-Lines: 63
> 
> Dear all,
> 
>   This is just to let you know that I have fixed a few problems with
> [the excellent] e2db, starting from the version distributed as
> EDIF_DB1_0_patches, which I obtained from LANL in early July 1997. I
> hope that these fixes don't conflict with or duplicate work that has
> been done elsewhere.
> 
>   Briefly, the changes are:
> 
> 1. check for a non-existent "bundle net" (which avoids a SEGV crash that
>    sometimes occurs - I'm not sure of the exact triggering circumstances
>    - when you have a bus that is connected to nothing)
> 
> 2. added attempt to use bus name whan fail to find a bus with the same
>    name as a hierarchical connector (avoids the need to give buses which
>    have bus taps the same name as their hierarchical connectors)
> 
> 3. ignored any part of a wire name that begins with a double underscore
>    (this appears to happen occasionally, with a trailing sequence number
>    following the double underscore; the wire name parsing routine
>    couldn't handle this, with the result that such wires were
>    essentially ignored)
> 
> 4. fixed EDIF file parse error which did not handle properties on wires
>    (the property name overwrote the wire name in the parse tree and thus
>    essentially caused the wire to be ignored)
> 
> The result is that some perplexing failures to make connections
> involving buses have been resolved (previously I had hit cases where
> apparently legal schematics were failing to result in the expected
> connections in the .db files; deleting and re-drawing wires at random
> would usually fix these problems but I would never know why; I suspect
> that this was usually the "double underscore" problem).
> 
>   There are some other things that would, I believe, be worth doing to
> e2db. For example:
> 
> 1. allow connections via hierarchical symbols to work via "sub-buses"
>    (e.g.  tracking bus[1] through bus[0-2] where bus[1-2] is tapped off;
>    this doesn't currently work because there is no wire named bus[1] in
>    the destination diagram; the logic to realize that bus[1] is a subset
>    of bus[1-2] needs to be added)
> 
> 2. allow "alias" symbols to work (these rename buses and would allow
>    "fat" buses in top-level diagrams to be broken out to "thin" buses in
>    lower-level diagrams; this would encourage more use of buses and,
>    therefore, fewer connections on top-level diagrams, and would support
>    a more "DFD-like" Capfast programming technique)
> 
> 3. detect more Capfast layout errors (e.g. an attempt to connect an OUT
>    field to a DOL field is almost certainly an error if the wire is not
>    also connected to a VAL field but does not result in an error;
>    instead, no connection is made in the .db file)
> 
>   I would be happy (at some stage) to make some or all of these changes
> (although aliases might be hard) but would like to get a feel for who
> else is interested and maybe even working in parallel on similar
> problems.
> 
>   A Happy New Year to all,
> 
>   William
> 
Hi William,
What a nice Christmas present, to come back and find that you had fixed
the bus problems with e2db.  The lab was closed from Christmas day until
today, so I didn`t get your first mail.  I don't think any one here has
been working on it recently.  I do think that we could eliminate the 
necessity of reading edb.def files by having e2db use the dbstatic lib routines
and I had been meaning to work on that, but haven't had time.

Could you put the results of your work on an ftp site where we can 
download it?  

Thanks for fixing so the problems,
Rozelle

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System