Tamás Nepusz
Tamás Nepusz
Okay, so if I understod correctly how it's supposed to work, commit 7b2aa7b2 does the trick and it is indeed nicer this way.
Ah, actually, I don't know. I wanted to be pedantic and get all the letters right, but maybe CRAN will complain. Shall we attempt a submission and then check whether...
The callbacks of the `bfs` / `dfs` functions seem to be fundamentally broken at the moment; in particular: * ~returning anything that is a non-logical value crashes R (and this...
`R_igraph_finalizer()` also takes care of hiding the progress bar in case it's still visible.
I don't know; I'm not that familiar with the R internals either, but calling `IGRAPH_FINALLY_FREE()` multiple times should be safe. The calling-igraph-functions-from-callbacks problem should be solved with the introduction of...
Addition: using `print(unclass(g))` inside the callback works, which points further towards the direction of `print()` (or, more precisely, `print.igraph`). Adding _another_ `print` after `print(g)` in the callback gets executed correctly,...
Okay, so it's simple and it will be solved in future versions when we switch the C core to 0.10, where we have the multi-level error stack available. The problem...
Makes sense to me. @vtraag ?
The proper solution to this issue would be to add an argument to `write_graphml()` in the Python layer that tells igraph how to serialize attribute types not supported by GraphML...
I think this has been done; a quick grep for `malloc` and `calloc` revealed only a few places where we use dynamic allocations and they make sense in that context.