[Librecores Discussion] Cocotb 1.1 is released!

Luke Kenneth Casson Leighton lkcl at lkcl.net
Thu Jan 24 09:10:11 CET 2019


On Thu, Jan 24, 2019 at 8:00 AM Stefan Wallentowitz
<stefan at wallentowitz.de> wrote:
>
> Hi Luke,
>
> On 24.01.2019 00:13, Luke Kenneth Casson Leighton wrote:
> >  super: i really like cocotb.  can i ask a favour?  there is an axi4
> > stream class that someone wrote: i have absolutely zero idea how to
> > use it: it came without an actual "usage" example.  the code is just
> > too obtuse and i simply do not have a handle on axi4 streams to be
> > able to even know what to begin to do.
> >
> >  if i had even one example that transferred just one byte in each
> > direction i would be able to adapt the example and use it to create a
> > unit test for a libre axi4 audio streaming PHY created recently.
>
> In my personal experience it makes sense to deep into cocotb a bit
> before starting. I know that this may sound a bit strange, but as you
> mention there is still some room for improvement in examples and
> documentation to get started more easily.

 appreciated, stefan: my expertise is specifically in
knowledge-derivation (reverse-engineering techniques using
epistemological analysis and in extreme cases bayes / demster-shafer
aka "black box" analysis), and i apply that skill to everything that i
do, that i have no prior knowledge of.  [weirdly / paradoxically /
fascinatingly, documentation is a *barrier* to the use of such
techniques!]

 basically what i can do is make very small incremental changes from a
*known-good* position.

unfortunately in this particular case i am confronted by more than two
unknowns, so there *is* no "known-good" to start from. consequently,
with too many unknowns i cannot make the "jump".

> In particular, the AXI modules already build on layers of abstraction
> (BusDrivers and Monitors) that make it much more convenient to use. I
> have recently (yesterday) written kind of similar modules that could
> also be used in your example. They still need to undergo major
> improvements to have the full power that the AXI modules have, but in
> their current state they are maybe an actual example of how a
> minimalistic usable cocotb module can look like:
> https://github.com/wallento/cocomod-fifointerface.

 unfortunately, it doesn't help in this particular case, as it's not
an example of how to *use* the axis.py code
[https://github.com/aoeldemann/cocotb/blob/master/axis.py]

 with no example of how to *use* that code, i cannot convert what does
not exist (in an incremental and iterative fashion) into something
that will test the libre MPAC PHY code.

 in talking with the developer, i asked if he had some example code,
and unfortunately did not receive a response.

l.


More information about the Discussion mailing list