Forum Replies Created
-
Posts
-
Yes, sorry about that. Qt is constantly changing and although they don't like to admit it, Qt5 broke a lot of things just like the Qt3-4 move did.
In fact I can't use Qt 5.4.1 on Windows or Mac with OpenGL. Have to stick with Qt 5.3 until they fix the bugs…
4.4.44 will have all platforms except CentOS 5 running Qt 5.x.
There are still some issues on Mac running Qt 5.4 (fixes some long standing bugs on Macs with Retina display, but 4.4.44 will default to non-openGL mode if it finds a Retina display, until this problem is resolved – it is being worked on.
You will need the version 4.4.43 which includes an env var "GLADE_READ_48_LIBS" which cabn be set to any value e.g yes.
The problem is that Qt did some undocumented changes to the serialisation of certain Qt classes, which broke compatibility between Qt 4.8.x (used on Linux and earlier versions of Windows) and Qt 5.x (used on more recent values of windows).
4.4.43 will have the option of reading this env var and I have tested it with on existing data and will solve the problem.
If you are running on Windows 32 /64 or Ubuntu64 this patch has already been applied and you can download the existing 4.4.42 version. Else wait for the fully tested version which should be released by 29/03/15.
Did you save the library from a diffferent machine type e.g. Linux and are now reading it from Windows? Or vice versa? There is a known bug here I am working on at the moment.
Failing that, can you send the lib to me at support @ peardrop . co . uk (less the spaces)? Or at the very least the library's glade.lib file.
Hi Steve,
Do you have seperate GDS layer numbers in your techfile? This is a must, else everything will get output to GDS layer 0 by default. You techfile should have entries like:
// Name Purpose gds_num gds_dtyp RGBA sel? vis? fillstyle linestyle
LAYER active drawing 2 0 (217,204,0,128) t t dots2 plain ;The 2nd argument is the layer name, the third the layer purpose (typically 'drawing' for GDS), then the 3rd and 4th are the mappings to the GDS layer numbers and datatypes. The remainder of the arguments are the layer colour in RGBA format, whether it can be selected by default, visible by default, what the fill pattern is and what the linestyle of the shape is.
Of course the problem could be in the other layout viewer, if you are not specifying GDS layer numbers to individual layers, I don't know.
Just to explain: GDS has only the concept of layer numbers and datatype numbers. In glade, the pair of these map to a specific layer and purpose which is what is shown in the LSW. You can double right-click on a layer in the LSW to show what the internal layer name/purpose is and what the GDS2 layer mapping is.
Beware: if you import a LEF file befiore a tech file, it will create its own layer name/purposes and of course it knows nothing about GDS2 mapping. Always load a Glade techfile before loading LEF (A cadence techfile will work too, as it allows reading a Cadence style layer map file for GDS reading).
regards
Keith
Mainly bugfixes, with a fix to the hierarchy browser for cells containing unbound instances, and some productivity improvements when creating instances in a library whose masters are in a different library – lib/cell/view names are remembered to make it easier to create multiple instances of the same master. Lastly the Cadence techfile parser has been improved to handle more recent versions like 616, and the gds layer map file reader now handles comment lines (i.e. lines beginning with a '#' character unlike the display.drf/tech.tf files which use the Skill ';' format)
Linux versions now use a later version of glew, so these libs have been updated.
Although there may be little obvious external changes to Glade, there have been some big refactoring work going on inside to make it more maintanable and more compatible to modern standards (e.g. C++ 11 when it is available on all platforms supported).
Also the version control has been changed to svn which is much nicer than most of the alternatives, at least for local networks.
The main improvement has been to the query dialog to allow changing common instance masters, for example. Any feedback appreciated.
Also I have been attempting to fix a bug in Edit->Stretch which I can't reproduce locally, but have made some changes that may prevent this bug ocurring again. Again any feedback appreciated.
In fact, changing common properties of instances and arrays was limited to just the origin. I am going to change the code so that inst/array objects can be changed as 'common' for the following:
– master lib name
– master cell name
– master view name
– orientation
– origin (not sure if this makes sense)
– magnification (ditto)This will need a bit of testing and will be in release 4.4.38, can't say exactly when that will be available though, hopefully before next weekend.
I'll have to try it out and look at the code. From the message it appears you are trying to change the layer of an unknown object, but that may be misleading.
regards
Keith
Hi Jofre,
Yes that seems to be correct. I'm away all this week, will check when I'm back. A workaround would be to select the object, then Edit->Move (not bindkeys etc). Enter F5, coords 0,0, then enter F5 again and the delta coords. The bug is in the mouseMoveEvent() code – I think I know what it is.
regards
Keith
Hi Folks,
I have had a couple of reports of people getting crashes when using>Stretch command. Unfortunately no-one's been able to give me an example and the actions needed to reproduce the problem, and I've been unable to reproduce the problem myself.
It only appears to happen in partial select mode when you are stretching a segment.
So, I need your help. If you come across this, and can send me a simple example ( a single shape would be ideal) and a log file and description if possible what you were doing when the crash occurred, it would me much appreciated – and then I can fix this bug.
Thanks,
Keith
Hi Jofre,
I'll update it and roll it into the next release. Thanks for pointing it out.btw the change of colour is because I changed the example.tch to have transparency (alpha channel) values of 128 rather than 255.
regards
Keith
Glade has moved on to use the later MSVC compilers – currently 2013 – which don't by default support Windows XP. Let's face it, XP was introduced in 2001, and Microsoft dropped support for bugfixes/security issues in April 2014.
However there are still a few people out there who are still running XP. In fact this site's web browser stats shows about 5.6% of visitors, compared to 5.8% using Win8.1. Well behind Windows 7, Linux and Mac OSX.
So I've attempted to build an XP-compatible version of glade. It works, at least on a XP virtual machine I've tested it on.
It does require the VC runtime redistributibles, however installing these on an old system can be troublesome. You probably don't even want to try anything running less that XP SP3.
Any feedback appreciated.
