[Keith]

Hi Einar,

the delimiter should not matter. Can you send me via email the schematic netlist you are trying to use and the extracted one? It will be easier to debug that way.

The error:

Graph "C:UserseinarDesktoptemptop.spc_flat": Fatal error at line 6: device 'XX1' not defined.

is a Gemini error, and it appears the netlist has not been flattenned correctly – Gemini only accepts flat netlists.

thanks

Keith

[Keith]

Hi Einar,

Glade expects a hierarcical (or flat) spice netlist with a 'wrapper' .subckt e.g.

.subckt top_cell_name CK Din rb Qout
XX1 CK Din rb Qout dffrbx2
.ends

to define the top cell. I agree that in theory this isn't needed – I checked in the HSPICE documentation – but it is just the way the tool works and many netlisters e.g. Cadence work in the same way.

I think you need to change it as above so it has a similar .subckt wrapper as the extracted CDL netlist. You would need to specify is in the LVS dialog as hierarchical, with the top cell as top_cell_name or whatever as in your extracted netlist.

Hope this helps

Keith

[Keith]

Please see e.g:

– user guide
https://www.peardrop.co.uk/glade/Glade.pdf

– example of integration with other tools
https://www.peardrop.co.uk/glade/abstract_poster.pdf

[Keith]

The Qt libraries on Windows have been updated to 5.2.1 which fixes some annoying bugs in the 5.1.1 libs. Specifically, when using e.g move or copy, the shortcut keys to mirror ('x' or 'y') or rotate ('r') did not work. More specifically, they did, but the event filter was getting passed the event twice and so cancelling out changes!

Same for e.g path entry with the 'u' and 'd' shortcuts to switch up or down a layer. And some other weird gui effects. Note on Linux/Mac I am still sticking to Qt 4.8.x as 5.x is either unreliable or will not compile on those platforms without a lot of work.

There have been other general fixes to resolve crashes in certain situations. If you find a crash, please do send the crash stack trace/logfile and (if possible) some data to reproduce it. I can fix 90% of bugs that have data with which I can reproduce; that probably goes down to about 20% for bugs with a crash dump/logfile and even less with just a description of the bug.

[Keith]

Mac and Linux ports will be upgraded to Qt 4.8.6 – note that 4.8.6 does NOT compile cleanly from source on OSX; you need at least the -no-phonon and perhaps the -no-webkit options. Fortunately Glade does not need these libraries.

I have yet to try to see if 5.3 is compilable or stable on all platforms (5.1 was definitely not)

[Keith]

The latest version, 4.4.13, is up on the website. It contains several bug fixes for crashes that have been reported, as well as several major updates to the DRC checks. Please update your version asap.

Also note that since Version 4.3.42 it is unnecessary to imstall a seperate verion of Python with Glade on Windows. The Windows version comes with a version of Python 2.7 DLL's and libs (you might still want to if you need python binaries like IDLE etc). This is automatically detected and used by the embedded Python interpreter.

Python is NOT installed on Linux / MAC OSX as these OS's normally come with Python installed.

[Keith]

Thanks for the example, this was a big help. I have also fixed the display group (item #2). It will be in version 4.4.11 due to be released today.

I've also improved the robustness of the parser somewhat.

[Keith]

#1 Mapping file has been done, but not tested.
#2 the code was actually originally written to handle display groups, and it handles them in the same way as packets in a Cadence techfile, but there is a subtle bug in that it expects the LayoutLayerTable lines to have the layer colors etc., else it ignores those lines. This should be easy to fix.

However, I do really need example .dsp / .tf files to test and make sure it works…

[Keith]

Hi Satya,

For #1, this should be straightforward. I'll make it read a gds layer map. assuming the same format as the Cadence one.

For #2, is this a new way of defining layers compared to previous?

If you could send and example display file/techfile this would help a lot.

thanks

Keith

[Keith]

If you find a bug or get a crash when using Glade, please report it, else it may not get fixed!

If you do hit a bug, here is what will help get it fixed:

1) The stack trace and log file produced when it crashes. The stack trace, as it tells me approximately where in the code it's crashing, and the log file, so I can see what events led up to the crash and info like what hardware you are using and what version of the code you are running.

2) Ideally also the example that caused the crash – e.g. any files (techfiles, GDS files, libraries/ LEF/DEF etc) that were loaded, any PCell python files used, any DRC/Extraction or other scripts used. Often the example can be reduced, for example a DRC problem may just require data for a single layer, and a few lines of the rules file that run the check that is failing.

I realise #2 may be more difficult due to confidentiality but a confidentiality agreement can be signed if required. Some bugs are impossible to reproduce without data, especially ones relating to boolean processing/DRC/extraction.

Many trivial bugs have been fixed over the last year or so with the help of end users. So please keep repoprting them to support@peardrop.co.uk. Oh and make sure you have downloaded and used the latest version of Glade as the problem may have already been fixed!

Thanks,

Keith

[Keith]

So far, the Linux ports for RHEL5/6 and Ubuntu 12 have been done and back-ported to Qt 4.8.5.

Ubuntu 10 is current work in progrsss as is Max OSX 10.9

Patches have been applied to the Linux 4.8 Qt distribution to avoid a polygon clipping problem that results in incorrect drawing of polygons using QPainter – this does not affect OpenGL drawing.

Edit: all the platforms except win32/64 are on Qt 4.8.5 with the clipping patch.

[Keith]

I had hoped that version 4.4.0 and above would use Qt 5.x. After all, it's been nearly released a year. However, there are some real problems:

– Functionality changes. These can be handled, though.
– Stability issues. In particular on Linux, it appears the port to use XCB has created many bugs with remote displays.
– Lack of support for existing platforms. Qt5 does not support RHEL5/CentOS5 which is the 4th most popular platform after win64/win32/Ubuntu64. This is really inexcusable.
– Failure to even compile on Mac OSX 10.9. Yes I guess it's not really their fault, as Apple have delivered an OS which is deficient in many ways with regards to back compatibility. But the beta was out a while, dear Qt folks…

So the bottom line is that I'm rolling back Linux ports to Qt 4.8 and the same to Mac OSX ports. The Windows Qt5 port I'll keep just to see what the stability/problems are like – so far I have had no reports of bugs. If you see any, let me know!

[Keith]

Hi Jofre,

Thanks for the testing!

Now I just need to get Qt 5.1.1 to compile on the Mac and that wil be added to the list.

regards

Keith

[Keith]

I finally got a 'definitive' answer from the Qt folks on the plugins. Basically there should not be a plugins directory at all; the subdirectories (imageformats, platforms, printsupport) should be in the same directory as the Glade executable.

I will be updating the 4.4.1 distributions to reflect this, but you can modify yours locally…

[Keith]

Few points:

Glade built for RHEL4 is no longer actively supported. I have an old build for this that works if you need it; however Qt stopped building on RHEL4 on version 4.8.0 so newer versions of the code cannot run on this old OS.

For RHEL5, same thing has happened with the move to Qt 5.x, as the Qt folks no longer support this OS. I can maintain a branch of the code that will run on Qt 4.8.x for a while but don't want to do this indefinitely.

RHEL6 is stable and supported as it Ubuntu 10.04. Ubuntu 12.04 is now supported and any feedback on problems with this build would be appreciated.

[Author]

Posts