Is getaddrinfo() supported in MorphOS, or is it still the old gethostbyname() we have to use?

https://blog.erratasec.com/2015/01/you-shouldnt-be-using-gethostbyname.html?m=1

At least Roadshow supports getaddrinfo()...

MorphOS supports bsdsocket.library API. This API defines gethostbyname() as the name resolution function.

Of course support will be added when we get around to implementing IPv6. Until then it isn't needed, nor does it offer any real benefits over gethostnyname().

PS. I disagree with the notion in that blog post. There's nothing wrong with gethostbyname(). Just because some API is old doesn't necessarily equal it being bad.


[ Edited by Piru 06.05.2020 - 18:48 ]

AFAIR Miami had its own extensions via miami.library, with POSIX socket functions. For example
inet_ntop, inet_aton, inet_pton, getaddrinfo/freeaddrinfo, etc.


(Sorry hit the edit button, not reply - Piru)

[ Edited by Piru 06.05.2020 - 18:59 ]

Quote:
True, but it is unlikely that you would be running Miami on MorphOS, even though it is possible with some care. However, I know no software that uses or requires these extensions.

Quote:
It should be easier to port software, even without the need for IPv6.


Quote:
I fully agree.

Quote:
GNU portability library (gnulib) provides a getaddinfo() wrapper function that uses gethostbyname() internally.

Either way, you're likely to run into far bigger problems than just getaddrinfo, at least you're doing proper port and not quick'n'dirty ixemul build.

[ Edited by Piru 06.05.2020 - 20:05 ]

Quote:

Thanks, I should check it out.

It will for sure be quick n' dirty.

Quote:

You are right. Likely not many software used miami.library for this, if any. Using Roadshow compatible LVOs in bsdsocket.library for these extensions would probably make more sense.

I have to use a wrapper for Oricutron and XRoar, so it's not an IPv6 issue... Gnulib is not always an option.

[ Editiert durch polluks 07.05.2020 - 09:52 ]

Will getaddrinfo() make an entry in 3.15/3.16 or is it way far down on the feature list for that to happen?

Adding that (even though it'd only do IPv4) would likely cause a chain-reaction when trying to compile POSIX stuff, enabling even more things that then wouldn't work. You can normally patch that function out in minutes.

Jacek is definitely correct here. Additionally, it'd also result in a class of executables that would no longer work with other bsdsocket.library implementations.

Meanwhile, what exactly would be the benefit of adding getaddrinfo()? The blog post referenced in the original post offers no plausible reason (that would be relevant for MorphOS as it is now).

Piru: I see, so it's better to wait for an updated bsdsocket.library then. The benefit: easier to port Linux C++ source code (in theory).