units; although that is a more proper usage, it also happens to
break a lot of existing scripts. Since virtually all of these
scripts start out with "snap internal" to force units, I have kept
this single usage, when "units" is still set to "default". That
keeps the older scripts from breaking, while hopefully scripts get
updated over time to use "units" instead of "snap". Note that there
are still ways to make this behavior break, such as running
"units microns" to set units and then sourcing a script that assumes
that "snap internal" will change the units, which will then not
happen. But batch scripts should be handled property with backwards-
compatibility mode.
TECHDEPBASE, as the layers below that cannot be directly painted;
this avoids putting things like "paint checkpaint" into the
output, which are harmless but useless.
".tcl" off of the name, so that the cell name remains the same, and
(2) it uses "units" instead of "snap", as the latter is no longer
used for setting the unit type.
differentiating between bar and square contacts on input. Also:
added handling of the "mask-hints" operator for GDS input as
well as GDS output. This is a bit more realistic now that the
mask hint properties are handled as scalable integers and not as
character strings.
grid rescaling while reading a GDS AREF record. The previous
commit is still valid and prevents excessive warnings and
unnecessary rescaling when a bogus value is given for an array
pitch when the array has only one row or one column. But the
scaling was still incorrect and would potentially cause an
error if an array pitch was at a finer scale than any data read
before it, forcing a valid grid rescaling. This now works
correctly.
data into an array reference in GDS; This may be ignored by most
tools (which are not trying to keep track of grid scaling) which
read the bogus value which then gets discarded because it is
unused (X pitch is irrelevant when there is only one column, and
Y pitch is irrelevant when there is only one row). But since
magic will try to interpret all input relative to the grid, a
bogus value can cause serious problems, and magic needs to actively
ignore the appropriate input in the case of rows = 1 or columns = 1.
to force internal units to be used; this usage is now deprecated
and needed to be changed to the "units" command. Also: Fixed a
separate issue with getting both "view bbox" and "view get" to
return values in the currently selected units. Also: Fixed an
issue with "property" when setting a property to a list passed as
a single value (e.g., "property FIXED_BBOX [box values]"), which
needed to null the string after each value.
converted to the new PropertyRecord type. Since this was in the
CIFPropRecordPath() subroutine, which has always been problematic
and generally useless, I took the opportunity to make that function
dependent on an option set by "gds savepaths true", with the option
set to "false" by default, so "path" properties will no longer be
created from GDS input unless the option is manually set to "true".
Also: Found that the returned boolean from "box exists" is
sometimes getting overwritten with a NULL result, causing a spurious
Tcl error messages to pop up. I could not find where this
overwriting happens (probably in a tag callback), so I just took the
hack approach of converting a NULL result to "false", which is always
safe to do.
magic to complain about reading GDS_START and GDS_END properties
from a .mag file, although its fallback method handles the properties
correctly, anyway, so there is no actual impact other than the
error message.
new properties, so created a "property compat" setting and made
it TRUE by default. This makes magic print all properties as
type "string" on output into a .mag file. Which is fine, since
it converts all values to the right type on input anyway. The
only thing that backwards-compatibility mode prevents is user-
defined properties that are not strings. That is a very rare case
and can be done by turning off comptability mode. Some time in
the future compatibility mode can be changed to be default false,
but there's probably no real need to do so.
not get transferred to a new node during a node merge; consequently,
the identity of the default substrate node gets lost, and node
capacitance to substrate ends up being placed on node "0"---
Probably okay for simulation, but not the expected result.
are handled. Properties were previously only character strings,
which had become cumbersome because properties were being used for
mask hints and bounding boxes, with the necessity of constantly
converting values from string to integer and back, which can cause
a performance impact as well as just being messy. The main difference
to the command is the addition of an optional first keyword argument
for the property type, which can be "string", "integer", "dimension",
or "double". All types except "string" can consist of multiple
values. Multiple values can be specified as separate arguments on
the command line, so that, for example, values of FIXED_BBOX or
MASKHINTS_* no longer need to be quoted. In addition, this completes
the handling of "units" implemented recently, as all properties of
the type "dimension" can be entered in the current units, will display
in the current units, and will scale with the database.
The code was not enforcing an order of precedence, which meant
that the lexigraphical order of EFHNBest() was able to override
the simpler case of one of the node names being a top level port.
reading the .sim file format to reading the .ext file format, which
is that .sim files use capacitance values of fF while .ext files
use capacitance values of (nominally) aF. So there was a conversion
factor of 1000 left over from the .sim file reading code that needed
to be removed, or else parasitic values come out to be 1000 times too
high.
the location of a node which was previously read as a port. But
the port location was not set when reading the port, only the
drivepoint, so the location is undefined and can cause a program
crash.
output which matches the code for the subcircuit definition output,
so that the pin order matches between the two. The previous fix
prevented pins from going missing in the subcircuit definition,
but didn't match that change in the instances. This should
resolve github issue #488.
this work is incomplete but cleans up the code a lot and removes
unused code and improves misleading variable and subroutine names,
as well as correcting an issue with "extresist" not recognizing
the "extract path" setting. There are no known impacts to the
operation of extresist itself.
code which are relevant and need to be included. Current state
is that hierarchical extresist "basically works" but entry and
exit points through the hierarchy are not being examined, so
results are currently based on port positions and not actual
connections. Also, proper distribution of coupling caps has not
yet been worked on.
Rebased, fixing merge conflict.
name with a trailing exclamation point over any other name. The
handling of global names might be properly implementable, but this
is not it, as it will favor a non-port default name or a subcircuit
name over a port.
(1) All parasitic extraction: The "defaultperimeter" and the
"defaultsideoverlap" commands were failing to exempt types other
than space from the list of edges from which fringing capacitance
is evaluated. This led to incorrectly considering the boundary
between types such as poly and nfet, or between metal1 and rm1,
to be sidewall areas. The "default" statements are supposed to
consider the most common usage, so the code has been changed to
make sure that only edges from material to space are considered.
In the rare case that a material-to-material edge in the same
plane should be considered a sidewall, the non-default statements
can be used instead.
(2) Hierarchical parasitic extraction: Magic was incorrectly
adding capacitances for subcells which had been output already
when handling subcircuit connections during "ext2spice". This
duplicate counting has been eliminated.
"extresist" command continues to work as before. However, the
method now reads from .ext files instead of .sim files, so
generating ".sim" and ".nodes" files is no longer necessary.
In addition, the core code of "extresist" was put directly into
ExtCell.c so that full R-C extraction can be run using
"extract do resistance" followed by "extract all", without
needing to run "extresist" at all other than to set parameters
(e.g., "extresist tolerance 10").
(Ctrl key + left mouse button) will start a wire at the current
cursor position with the wire values set by "wire type" and
reported by "wire values". So "wire type metal1 0.28um" will
always start a 0.28um wide wire of metal1 regardless of what is
present at the cursor location.
parser that allows simple expressions to be entered for dimensions,
such as "2um + 2um" or even mixtures of units like "3um + 200i".
This feature is currently experimental.
and area) so that they are consistent across commands. The default
behavior remains the same, for backwards compatibility. However, a
new "units" command has been added, so that "units microns" results
in measurements always being displayed in microns, with choice of
that or "internal", "lambda", or "grid". The units themselves may
be printed (for interactive use) or not (for scripted use). The
use of "units" is independent of "snap", after overriding the
default behavior, so that units parsed on the command line are
interpreted according to "units", not to "snap".
(1) Fixed an error that was introduced in version 8.3.590 with
a patch that should have been applied only for the case of
BJT devices, and not for MOSFETs. The patch will cause
devices generated by "device mosfet" or "device asymmetric"
to be read incorrectly from a .ext file during "ext2spice".
(2) Fixed an error in the tech file reading, where using CDL
parameters on a capacitor device would cause the tech file
loader to print an error message. The parsing was correct
and only the message should not have been printed.
(3) Added a new feature with the new command option "extract
do unique". This replaces the "extract unique" command by
running the same code within the extraction, but has the
additional effect of reverting the label changes afterward.
This prevents the user from inadvertently writing the
altered labels back to the database file.
R. Timothy Edwards
Daniel Hwang