Make "last" (and similar variables) read-only

[ChrisJefferson]

This stops users writing to last, time, and some other variables, by making them read-only.

It adds a method to write to these variables, UpdateStat, mainly for use in demo.g.

I realise this might break some code. However, I did have someone had a had-to-track down bug where they were using last as a variable themselves, and were then confused when they got stupid answers out, so I think it is worth changing.

[fingolfin]

Thank you, this seems like a very sensible thing to me. However, I have various nitpicks.

[fingolfin]

Since you are touching this code anyway, perhaps you'd also be willing to teach it to update memory_allocated, too? Also perfectly fine if you don't want to, of course!

[ChrisJefferson]

I've never actually used this code, and don't even know how it should be used, so I'd prefer to leave it. I've just noticed it has no tests, no documentation, and isn't linked to from anywhere?

[fingolfin]

It is used like this: you take a file and put a bunch of inputs in it, each on its own line. Then you run Demonstration on this; whenever you press a key, the next input line is read and executed. The result then looks something like this (where each demo> appeared only after a key was pressed):

gap> Demonstration("demo.g");

Start of demonstration.

demo> 1+1;
2
demo> Factorial(10);
3628800
demo>
End of demonstration.

gap>

Updating last, last2, etc. only is of interest if the demo code references them. So if one wants to write a demo that showcases those, and also memory_allocated, then it'd be nice to have memory_allocated. But this is nothing this PR needs to do.

I've long wanted to give this code some love, and improve it; e.g. instead of always going to the next command upon any key press, I think only "return" on an empty line should do that; otherwise, it'd be nice if one could show the next command w/o executing it, and also edit it; and other stuff. I think other people implemented such demo modes for other languages, and they can be very useful for classes, presentations during talks etc.

[ChrisJefferson]

I think all issues are now fixed.

[codecov]

Codecov Report

Merging #3226 into master will decrease coverage by <.01%.
The diff coverage is 62.16%.

@@            Coverage Diff             @@
##           master    #3226      +/-   ##
==========================================
- Coverage   84.75%   84.75%   -0.01%     
==========================================
  Files         689      689              
  Lines      336391   336411      +20     
==========================================
+ Hits       285107   285116       +9     
- Misses      51284    51295      +11

Impacted Files	Coverage Δ
src/gvars.h	100% <ø> (ø)	⬆️
lib/init.g	83.3% <ø> (-0.04%)	⬇️
src/streams.c	72.25% <100%> (-0.09%)	⬇️
src/gvars.c	86.04% <100%> (+0.05%)	⬆️
src/gap.c	81.14% <53.33%> (-1.42%)	⬇️
src/hpc/threadapi.c	46.82% <0%> (+0.04%)	⬆️
src/objset.c	85.35% <0%> (+0.22%)	⬆️

[coveralls]

Coverage increased (+0.03%) to 84.677% when pulling c5b6027 on ChrisJefferson:protect-last into 211faf8 on gap-system:master.

[fingolfin]

Perfect, thanks! I am waiting with the merge a bit, so that the PR adding C++ coverage support can finish first, as otherwise codecov will probably once again be super confused about what it displays sigh

[ChrisJefferson]

Fixes #338 ( mentioned here so it will get closed)

[fingolfin]

That doesn't work, you have to add the "fixes" to the PR description