Not sure why providing a (slightly outdated) package is a disservice. Ubuntu 20.04 provides the same version https://packages.ubuntu.com/focal/ns3 and last commit is ~1 year old.
The package is non trivial, non broken and it's not like having an older version is a security hazard either.
I pushed this package because I am the https://github.com/direct-code-execution/ns-3-dce maintainer and I thought ns-3 would have a lot to gain using nix since their custom package management is broken, with many dependencies. I never had the time to properly disseminate the nix knowledge to the ns3 community sadly and the odds of me doing so are decreasing.
I think we could move this to an ns-3 flake when flakes get available ?

Hi,

I see it as a disservice because using an old ns-3 version when a new one is out means accepting simulation bugs which lead to wrong or biased simulation results. One could argue every new version also introduce bugs, but, in my book (and I might be wrong), there is not a single advantage to keep using an old ns-3 version.

Indeed, the ns-3 build system, which heavily relies on waf, seems complex and could be seen as broken. But to me, it seems nobody is /actually/ supporting this package and using it at the same time. Concerning ns-3-dce, I was told it was not really maintained anymore ( https://gitlab.com/nsnam/ns-3-dev/-/issues/91 ) so I didn't expect it to be supported anyway.

But I get your point: if no one works on this package, then its presence is not really diluting the workforce and its presence is not really a problem.

If I update the package to ns3.32 does that address your concern ? It should be just a matter of changing version+hash (same aomount of work than removing it :) ).
I've packaged DCE for nix but not upstreamed it yet.
One problem I have with DCE is that I wanted to have it compile on nix with the kernel associated with it since the main interest of DCE is to run determinisitc experimentation with a linux kernel. But this is not straightforward and also DCE only supports older linux (up to 4.4 IIRC). The other maintainer (Hajime Tazaki) is currently working towards upstreaming the Linux Kernel Library in linux, and I expect DCE to be "revived" once that's done. But as often with this kind of project (academic/teams of ~1person), it takes a long time :'(

Well, if you can update it, why not (my main point is: do you think you have time to keep it up to date in the future ?).
But since a few versions, ns-3 uses waf and python3 specific features in its wscript files, so, I guess you'll have to work around #69305 … Good luck for DCE, I really hope you'll be able to get where you want to ! (and I totally get the comment about the academic side).

In anyway, thanks,

it got updated in hte linked PR so closing.

Nixpkgs master is the most up to date package source available according to repology. I don't think it adds anything if we remove slightly outdated packages which are almost trivial to update. If such a case rises again feel free to ping me and I maintain the package.