[Oberon] Lola TS [WAS Re: Assumed issue in the Lola-2 compiler]

rochus.keller at bluewin.ch rochus.keller at bluewin.ch
Thu Apr 25 15:10:51 CEST 2019


@ Paul Reed

Thanks for your feedback.

>> I agree it's tempting but I was rather glad I didn't change the TS to 
>> generate inline 'z expressions, when I came across a toolchain (Clifford 
>> Wolf's IceStorm tools for the Lattice iCE40) which didn't implement 
>> tri-state properly from the Verilog.

I hoped it was fixed in nextpnr. The issue was primarily that arachne-pnr couldn't handle the _TBUF_ cells generated by yosys. I will check that. Of course we can implement our own module but I think it's a pity if the compiler assumes an implicit module dependency.

>> Delete the following IF in LSC.Statement
>>   IF z # x THEN x.a := z.b; x.val := 1 END

Please also note that Jörg Straube informed me that his fix is not the final solution and there are still other issues in the Oberon code of the Lola compiler. I started to develop my own Lola-2 compiler and will also implement a Verilog generator (not only a ILANG generator) to be less dependent on the Oberon version.

Best
R.K.


More information about the Oberon mailing list