[Keith]

There was a bug in the initial upload of this version whereby you couldn't open older libraries generated in previous releases. So please download the latest version compiled yesterday.

thanks

Keith

[Keith]

This is now implemented in 4.4.34, on the website for most platforms…

[Keith]

Hi Paco,

I'm not sure how this helps. You have to go to the selected object to deselect it, would it not be better not to select it in the first place e.g. by restricting the selectable layers? Or just use ctrl-LMB to delesect the particular object?

But yes, this is probably not too difficult.

regards

Keith

[Keith]

I think I know why this is occurring. Even though the viainsts do not get merged, they were getting deleted but the spatial tree wasn't being updated, so this caused the crash on Linux (for some reason the problem does not happen on Windows).

I've changed the code so it now merges shapes layer by layer. Note that it still will not merge viaInsts / MPPs (or in fact anything that is not a shape). If you want to merge these you will have to select them first and use Edit->Convert to Polygon to smash them before merging.

[Keith]

Hi Paco,

Layer by layer merge should be possible. I'm not sure keeping the existing functionality (merge to the first selected object layer) really made sense but then merge selected was really only intended to merge shapes on the same layer.

As for a crash trying to merge a viaInst and a rectangle, I tried creating a viaInst (m1_m2 from example.tch) and a rectangle on m1, selected them both and hit merge selected. I get the message:

>>> ui().editMergeSelected()

>>> # ERROR: Can't merge a VIAINST

and indeed in the code viaInsts aren't considered in the merge selected function.

[Keith]

Hi Paco,

You can now. 4.4.33 is up on the website. Just need to run the Mac port and then all platforms will be there…

regards

Keith

[Keith]

Hi Paco,

It wasn't so difficult after all. It's all working now, with a popup F3 dialog to change the ruler snap angle (this does not change the general snap angle in the display options dialog), and a checkbox to invert the minor/major ticks and numbering.

It'll be in 4.4.33.

regards

Keith

[Keith]

Hi Paco,

Actually I think I've figured out what might be wrong here. The Edit->Hierarchy->Flatten command was supposed to take a selected set of instances and flatten them; originally it was expected that you selected an instance and hit 'flatten' and it flattened that instance. In the code, it checked if the first selected object was an instance, otherwise it tried to find one under the current cursor position.

Now if you do a select all, then you may get objects that are not instances in the selected set, and so it will try and find an object under the cursor (wherever that is) that is an instance, and if that fails, then you get the error.

Saving the library and reopening it will reorder the objects. Which may explain why it works if you save, close and re-open.

The fix, to be in the next release, will be to check if any object in the selected set is an instance, and if so flatten all selected instances.

regards

Keith

[Keith]

#1 is done and will be in the next release.
#2 is more tricky. It involves a rewrite of the whole ruler drawing code.

[Keith]

Hi Paco,

This is the expected behaviour. Edit->Hierarchy->Flatten takes a selected set of instances and flattens them (i.e. promotes their contents to the current cellView).

If you want to 'flatten' a MPP, you can select it and use Edit->Convert to Polygon.

For the second issue, can you send me a testcase I can reproduce? I tried creating two subcells and placed instances in a top cell. I copied the instances and used Flatten on all of them, and they all flattenned as expected…

regards

Keith

[Keith]

Hi Paco,

That should be doable.

regards

Keith

[Keith]

Note that labels in Glade are the same as GDS2 texts and have no physical shape. Elsewhere on this forum you can find a script another user created to create physical shaped labels…

[Keith]

It puts the files in the directory glade was invoked from, not the directory the glade executable is in.

So if you run from a cmd line, it will put the files in your current working directory if the env var is not set. If you double click on the icon, I'm not sure where the executable is considered to have been invoked from – and also your ~/.bashrc env vars are not considered in this case.

[Keith]

As per the documentation, glade writes temporary edge files to the directory where the glade executable is run. If it can't open a file in write mode in this directory, then you will get this error.

However you can change this by setting the env var GLADE_DRC_WORK_DIR to put them in some writeable temp directory.

The message '[file ../../glade/libcpp/cpp_stream.cpp line 263]
' does not refer to a file on your system, rather it's indicating the program line the error occurred in.

[Keith]

I'm puzzled by how you get to create a library with anything other that the defaults which are set in the constructor of the library class to 1000 and 1.0e-9. Maybe you can give an example.

Actually the dbu value is only used for GDS2 input/output; it's purely there for historical reasons and is not actually used anywhere else. User units are always assumed to be in microns.

[Author]

Posts