Archon explicitly allows constants (c.f. parameters) to be floating point values to represent voltages, but wdl does not allow this

[weatherhead99]

In many places in the archon manual (e.g. pages 58 and 59) it states that both timing state definitions and slew rate definitions may use one of the defined timing constants as their values. These values are expressed in volts per us or Volts, and hence are floating point.

In fact this is one of the only very strong uses of the "constants" as opposed to the "parameters" in the archon timing core.

WDl currently does not support defining constants as floating point values due to the nature of the regex that lexes them, for example the following code in wdl is fine:

const my_const = 1

whereas the following is now

const my_const = 1.0

the error is thrown from the python regex parsing module.

Not sure whether this is quite a feature request or a bug really. I know that e.g. Tim never used constants but I often do for the reason that they are much easier to query than parsing an entire config to get out what the slew rate is, and it allows easier changing / tuning of slew rates and clock voltages at runtime.

[weatherhead99]

ok I don't know if anyone is looking at these @prkrtg @scizen9 @astronomerdave but I now think this is perhaps the most important problem in wdl as it stands. Everything else can be got around with (sometimes cumbersome) workarounds. But the lack of floating point constants effectively kills off any way to programatically set clock voltages except for really brutally hideous sed + shell script + reload the entire acf hacks. With floating point constants I can rewrite a single archon config line live, know where it is, and programmatically explore clock voltages which would be defined in these constants. Without them I am left to write that value in the appropriate place in (uncompiled) wdl, recompile the wdl, reload the entire config. Very flaky and annoying to do.

[scizen9]

Unfortunately, I cannot assign people to this issue, but please assign @astronomerdave and @prkrtg. You can also assign me, but I'm leaving the country for two weeks. I'll notify them on slack as well before I go. I hope this can be fixed. Sounds like a right pain.

[prkrtg]

#34

[astronomerdave]

@weatherhead99 Is it okay if constants are always floating point? I.E., if you enter

const foo=1
const bar=2.2

is it acceptable if the resultant ACF has

CONSTANT0="foo=1.000000"
CONSTANT1="bar=2.200000"