[Keith]

There are 2 things that need fixing here.
[ul]
[li]You need to open an existing library, not create a new one.[/li]
[li]You need to open the cell in 'a' mode, not 'w' mode, as this requires that the cell does not exist.[/li]
[/ul]
so e.g.

Code:

lib = getLibByName("foolib")
cv = lib.dbOpenCellView("foocell", "layout", 'a')

[Keith]

I'm aware of a bug that causes a crash when running boolean operations that involve processing shapes with holes, or generating shapes that contain holes.

I have some idea on how this may be caused, unfortunately so far no-one's been able to provide me with an example to debug and fix the problem. If you have such an example – preferably as simple as possible – please contact via the support email in the Help->About dialog.

Thanks,

Keith

[Keith]

If you have problems getting Glade to run on Linux, it may be because certain dependencies for the Qt plugins are not being met. For example libfontconfig.

One way of debugging this is to set:

QT_DEBUG_PLUGINS=1

in your shell before launching Glade. It will print out a mass of info on the plugins its trying to load, and any errors found.

btw Glade runs quite well on Windows 10 using a bash shell – the latest win10 version supports Ubuntu 16.04. You need an X server like Xming running on Windows for the graphics to work.

[Keith]

A nasty bug has been fixed in 4.6.4 that caused some shortcut keys (bindkeys) not to work on the Mac. This is actually a bug in Qt itself… the Qt folks have been notified.

[Keith]

This was fixed in 4.5.32, or is it still an issue for you?

[Keith]

Not sure how that can happen then. Do you have a reproducible testcase?

[Keith]

I don't understand. Are you saying e.g. you open a library, and you have a cell with a shape on (say) poly, and you get a crash, then reopen the library and the shape is now on (say) metal?

It may be you used different techfiles/library (glade.lib) files, and the layer number/order is different in which case you can run into problems, as layers are internally stored by a number – so if the mapping changes, then shapes will appear to change layer.

In general, don't read in a techfile (or a Cadence techfile) onto an existing library, or import techfiles multiple times, it's not required 99% of the time and can cause several side effects.

[Keith]

No sorry there is nothing like that. In theory creating multiple paths is straightforward, in practise it would be tricky for anything other than a single layer with no via up/down like the current path cmd.

[Keith]

you're missing flags. Add a 0 after length, before msg.

[Keith]

Currently it does not generate pin shapes. It generates instances of devices in the schematic using the instance properties to build the pcells. And the connectivity is created also so you can see what is connected to what via flightlines / cross probing.

[Keith]

In 4.5.31, geomSpace() will have a new format for wide metal checking:

geomSpace(layer, sep, width, length, flags=0, message="")
geomSpace2(layer, sep, width, length, flags=0, message="")

This will check the layer for spacing violations less than sep, for shape(s) wider than width and parallel run length greater than length.

So in the perpendicular track case, set length to a value less than the projecting length of the track, then these short edge tracks will not be flagged as in error.

[Keith]

Quote:

PS: I tried to send the attachment through a email to support@peardrop.co.uk, But I get a mail delivery failure stating "Your message wasn't delivered to support@peardop.co.uk because the domain peardop.co.uk couldn't be found. Check for typos or unnecessary spaces and try again"

its telling you that there is a typo in the name

[Keith]

OK, I can see what you're doing.

The issue is how do you define the 'width' of a shape (in this case a rectangle). THe check looks at the edges of two shapes, and finds outer edges that project onto each other. Then for those edges, the 'width' is the distance to an opposite parallel edge of the same shape.

This gives the expected result when two tracks that are long and narrow are parallel. When they are perpendicular however, the 'width' of one by definition is the longer length of the shape.

One possibility is to define 'width' as the shorter of the width/length of the track. Is this what you expect?

[Keith]

I don't know why the forum won't allow attachments. Please send via support@peardop.co.uk instead.

[Keith]

can you give me the exact command line you are using?

[Author]

Posts