character literals / enumeration literals

[imec-myhdl]

character literals ('0', '1', 'Z' etc) are converted to an integer in line 176 of src/vhdlConvertor/literalParser.cpp by subtracting the value '0' This produces 'weird' results for anything else than '0' and '1'. for example std_logic is predefined enumeration('U', 'X', '0', '1', 'Z', 'W', 'L', 'H', '-') Also the bit width is set to 8 (in line 168 of the same file).

Solving is probably easiest by doing a 2-step (visitor) approach: during the first visit do type annotation (like in myhdl) where signal/variable types/widths/initvalues/signed are resolved (in the myhdl case relevant info is stored in attributes of the nodes, but that is dynamic, so might not be the best approach in cpp, During the second visit the type is known, and conversion is relatively simple.

[Nic30]

• Yes, current code is temporary solution, I will fix it tonight.

Problems:

• Char literals can represent 1 to 8 bits depending on context.
• If we want to resolve the type of the signal in simple case analysing the current file is enough but in generic case it is required to parse all librarires and and other project files. From this reason it would be very good to avoid such a analysis in this library (Because this is next level in parsing and this library is only lower level).

I will change mentioned code to represent char literals with bitstring instead (so the value is always correct). And I will add a flag which signalizes that this was char literal (so it is easy to convert it back without code analysis).