Re: [myhdl-list] Prerequisites to run the MyHDL regression test
Brought to you by:
jandecaluwe
From: Christopher F. <chr...@gm...> - 2015-03-01 22:53:44
|
On 3/1/15 7:50 AM, Josy Boelen wrote: > >> You should be fine with the Altera Modelsim then, >> you can run the `make vcom`. Are you able to run >> `make vcom` with git-bash (you might have to add >> the Altera path) in conversion/general? >> >> Regards, >> Chris >> > Chris, > > The `make vcom` doesn't work as git-bash expects the .exe extension to > find `py.test` . Hmm, that is odd. On the Windows system I am using it works - but I don't recall if I did something custom during setup ... On the Windows system I am using the Enthought dist (a Canopy version). I am fairly sure pytest is installed with the distro, if it wasn't I probably installed via `pip`. > If I give the command `py.test.exe vcom.py test_*.py >utest.log' it seems > to run, be it with a lot of failures. Ok, this seems to be an issue because I see the same thing but I have no idea why (right now). A quick look it seems the VHDL myhdl pck is empty, it might be an issue with the latest directory selection addition. > But I have `myhdl-0-9-dev` sitting alongside my local `myhdl' and of > course the test is using this `myhdl`-tree, and I have a few experimental > changes: I see all the errors, on Windows only, with no mods. > like instead of `if bool( sig ) then` I made toVHDL() output`if (sig = > '1') then`. And the first test fails on that as it has this in the > converted putput: ` a <= (not (a = '1'))(0);` where the unmodified myhdl > would have produced ` a <= (not bool(a))(0);`, which I don't quite > understand either. > > The main observation is that I'll have to set up that virtual Linux > machine anyway. Possibly but in general I think we want to make sure things work on most major systems Python support? It is odd, from my background most (not all) FPGA developers work on Windows and most ASIC devs work on Linux. Regards, Chris |