mipsel-linux-uclib: delete ports

[mojca (Mojca Miklavec)]

The ports for mipsel-linux-uclib:

• mipsel-linux-binutils
• mipsel-linux-gcc34
• mipsel-linux-kernel-headers
• mipsel-linux-uclib
• mipsel-linux-uclib-headers

have last been updated 12 years ago (ignoring cosmetic fixes), we had two bug reports (which I understand as if the ports were mostly useless, but I may be wrong), the last one four years ago (#38135, #43988). Homepage that some of the ports point to no longer exist (​http://mipsel-linux-binutils.sourceforge.net/). The buildbot-generated packages from 2016 apparently still exist though, so maybe some of those ports are still able to build, but that doesn't change the perspective too much.

Unless someone knows the purpose of these ports and is willing to update them, I see no reason to keep shipping gcc 3.4 etc. Building mipsel-linux-uclib with the crossbinutils PortGroup alone worked, but I have no clue if the result is usable.

Should anyone know the reason for not deleting these ports, please speak now.

[jmroot (Joshua Root)]

I don't use them, but if that was the threshold for deletion we'd delete most of the ports. I'm sure most of our users don't develop for MIPS Linux but it's entirely possible some do, and if they do then in all likelihood they don't know this ticket exists. Debian has these packages:

​https://packages.debian.org/buster/gcc-7-mipsel-linux-gnu ​https://packages.debian.org/buster/binutils-mipsel-linux-gnu

GCC 3.4 is definitely old and should go. Current GCC and binutils versions appear to support mipsel-linux as a target. uClibc can also be updated.

[mojca (Mojca Miklavec)]

But we do in fact have way too many broken and unmaintained ports, many of which might not have worked for ages.

I've seen those packages in Debian, but gcc-7 and gcc34 are two entirely different things. Also, there's probably difference between mipsel-linux-gnu and mipsel-linux-uclib. I'm not even sure which one we are supposed to support. Both? I don't think we currently support mipsel-linux-gnu.

If you are willing to take a look at updating them, that would be great. I assume that if there was sufficient interest, we would at least get one single request for updating the compiler. The version has not been touched for the last 12 years.

I would say that we should either bring the ports up to date and make them maintainable, or put them to rest. Keeping unmaintainable patches for gcc 3.4 doesn't really serve anyone.

[pmetzger (Perry E. Metzger)]

I think Mojca is right. If there's no upstream, and no known purpose to the thing, and no maintainer, and it's partially, broken, and it's for an ancient version of gcc, and no updates in 12 years, it probably should go.

[jmroot (Joshua Root)]

If the ports don't build (anywhere) and there's no maintainer and no one else willing to fix them, then yes, they should be deleted. I'm just not sure how much good it does to ask the small number of people Cc'd on this ticket if there's any interest. (The purpose of the ports is clear; they're for MIPS cross-development. Upstream is also clear, it's now mainline gcc and binutils.)

[mojca (Mojca Miklavec)]

If there is full support upstream, then creating four ports for binutils and gcc for potentially both architectures should be trivial and we should be able to get rid of all the patches. I can do that, but have no clue if the resulting packages will be even functional as I have no clue how to test and I don't know how to judge whether 100% of our patches are obsolete. I'm pretty sure that the current state of ports doesn't do any good to anyone, so I would prefer having at least one user who is able to test the software to give a green light about its functionality.

And someone needs to bring those ports up to date, of course. If we can update them, then sure, we can keep them.

[mojca (Mojca Miklavec)]

In b341a7c3770cb00226724c4282421c778641ae78/macports-ports:

mipsel-linux*: delete ports

Closes: #56108