Consider func and and sideEffect free

[arkanoid87]

I was testing measuremancer on my fully "unchained" and "sideEffect" free lib, but I've found out that measuremancer calls imply Side effects

Hint: 'computeW0Deadh' calls `.sideEffect` '-'
>> /home/arkanoid/.nimble/pkgs/measuremancer-0.2.1/measuremancer.nim(216, 6) Hint: '-' called by 'computeW0Deadh'

[Vindaar]

The reason is that we have a global ID counter in the module to keep track of the symbols. I'll think about whether I can replace that either by a fully CT counter or some other solution. But I wouldn't bet on it.

[Vindaar]

Well, I have a CT variant working locally. But that requires me to turn ± and measurement into macros (so that I can register every call to those procedures and increase the ID counter at CT.

Two things:

• I'm not sure if I break some use cases (essentially every code path gets the same IDs. It should be fine I guess, for same inputs it's the same variables. For other inputs the value / uncertainty pairs are different. But it's quite possible I miss something
• turning the basic API into a macro for this isn't great

[arkanoid87]

If side effects are a requirement, I prefer an easier to maintain measurement package than a convoluted compile time thing to just make it work.

I will convert my code to proc instead. The pro are far larger than the cons

[Vindaar]

Yeah, I tend to agree.

I've pushed the (not cleaned up) code here https://github.com/SciNim/measuremancer/tree/CTcounter if you want to look at it / play around with it.