feat: remove deprecated functions before 1.0

[Antonov548]

Deprecated

• [x] igraph_bipartite_game() — see #630 - not use at all

Updated

• [x] igraph_hub_score() - replaced with igraph_hub_and_authority_scores(), see https://github.com/igraph/rigraph/issues/1177#issuecomment-1917980983

• [x] igraph_authority_score() - replaced with igraph_hub_and_authority_scores(), see https://github.com/igraph/rigraph/issues/1177#issuecomment-1917980983

• [x] igraph_lattice() — replace with igraph_square_lattice(), see #994

• [x] igraph_laplacian() — replace by igraph_get_laplacian()

• [x] igraph_erdos_renyi_game() — not used in R.

• [x] igraph_random_edge_walk() — functionality merged into igraph_random_walk().

  R side

• [x] hub_score() - added hits_scores. hub_score is deprecated and use hits_scores under the hood.

• [x] authority_score() - added hits_scores. authority_score is deprecated and use hits_scores under the hood.

• [x] make_lattice() - use square_lattice(). circular parameter is deprecated. Added periodic parameter. periodic should be logical vector with same length as dim or single value which will be converted to vector.

• [x] laplacian_matrix - use generated get_laplacian_sparse_impl and get_laplacian_impl instead of R_igraph_laplacian. normalized argument is deprecated with new normalization argument.

  • [x] igraph_erdos_renyi_game() — not used in R.
  • [x] igraph_random_edge_walk() — updated random_edge_walk to use random_walk_impl with edges return.

[krlmlr]

I don't see a caching step, can you please be more specific?

[Antonov548]

I don't see a caching step, can you please be more specific?

I mean this caches. I already had once the problem with them - https://github.com/igraph/rigraph/actions/caches. But as I know we don't have easy way to clear all of them.

[Antonov548]

@krlmlr the issue wasn't the cache. I had some troubles with search in VS code, so that I didn't see another usage of removed function. But it's good right now and maybe it make sense to run check for packages once again.

[krlmlr]

This looks green. Running revdepchecks.

[krlmlr]

Looks good enough to me?

[Antonov548]

Looks good enough to me?

Also looks good to me. Thanks for checking it.

[szhorvat]

I'm on my phone and can't look properly, but this caught my eye:

Error in igraph::graph.lattice(length(subwords) + 1, directed = TRUE) : 
  argument "circular" is missing, with no default

This is changing a deprecated function, which is perhaps not a good idea?

The circular / periodic parameters are otherwise not mandatory, right?

[szhorvat]

The netropy issue is not our problem.

Did anyone look and check that the tidygraph one also isn't?

[krlmlr]

tidygraph is present in main, I think this was also something that we can/should fix, in addition to the circular argument.

@maelle: can you please take a closer look?

This isn't meant to be a blocker for this PR, just something that we want to do before release.

[szhorvat]

This isn't meant to be a blocker for this PR, just something that we want to do before release.

Absolutely!

[szhorvat]

make_lattice should also be renamed make_square_lattice to make room for make_hex_lattice() and make_tri_lattice() (or whatever you want to name them). https://github.com/igraph/rigraph/issues/994

[krlmlr]

Thanks!