Add sugar to set fields with PointerWrapper

[lcw]

This enables the setting of the user_pointer field of a p4est. This is useful to access julia data in p4est's callback functions. For example:

conn = p4est_connectivity_new_brick(2, 2, 0, 0)
p4est = p4est_new_ext(MPI.COMM_WORLD, conn, 0, 2, 0, 0, C_NULL, C_NULL)
pw = PointerWrapper(p4est)
pw.user_pointer = pointer_from_objref(Ref((3,4)))

[codecov]

Codecov Report

❗ No coverage uploaded for pull request base (main@177d26e). Click here to learn what that means.
The diff coverage is 80.00%.

❗ Current head 1f5b561 differs from pull request most recent head b00fec5. Consider uploading reports for the commit b00fec5 to get more accurate results

@@           Coverage Diff           @@
##             main      #77   +/-   ##
=======================================
  Coverage        ?   15.63%           
=======================================
  Files           ?        3           
  Lines           ?     1516           
  Branches        ?        0           
=======================================
  Hits            ?      237           
  Misses          ?     1279           
  Partials        ?        0           

Flag	Coverage Δ
unittests	15.63% <80.00%> (?)

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
src/pointerwrappers.jl	86.36% <80.00%> (ø)

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

[ranocha]

Thanks a lot! Having setproperty! as analog of getproperty is a nice idea.

@JoshuaLampert Could you please please have a look at this and merge it if it looks good to you?

[sloede]

Looks good from my side, thanks for the contribution!

I wonder if it would be necessary to also add one or two examples to the docs for correctly using this feature. AFAIU, one has to GC-preserve any variables before getting their pointer, at least if the variable is not used afterwards anymore, such that the memory address remains valid. This is of course not an issue of PW but if Julia itself, however, since we expose some very low level stuff with a nice convenience layer, less experienced users might fall into hard-to-debug traps.

I am thinking of maybe one example for a persistent variable to show general usage, and maybe one for a local variable where GC preserve is necessary for defined behavior.

What do the others think?

[JoshuaLampert]

When this is documented this LGTM. Thanks!

[lcw]

Okay, I added a minimal example.

[JoshuaLampert]

I just realized: Does this also work with these weird structs as in #72 or should we also take a similar approach as in #74?

[ranocha]

I just realized: Does this also work with these weird structs as in #72 or should we also take a similar approach as in #74?

Good point. Now that #74 is merged, I think we can proceed like this.

1. Merge this PR. It already gives a nice new feature, so we can go ahead.
2. You can check it afterwards and make a PR if required.