[wip] gcc 4.9.4 package

[th0ma7]

Motivation: Intention is to provide an initial gcc package in hope of eventually providing a newer version of the full gcc suite. This provides other opportunities such as replacing entware gcc and ability to build perl bindings using cpan requiring a C compiler.
Linked issues:

Checklist

• Build rule all-supported completed successfully
• Package upgrade completed successfully
• New installation of package completed successfully

Note that due to the large amount of resource files needing to be provided I've create a new PLIST.auto functionality. I allows using an empty PLIST.auto in cross/<package> where, when captured, the spksrc.plist.mk will look at files provided by $(WORK_DIR)/$(PACK_NAME).plist using file to determine its type and automatically create PLIST entries for it.

This should be quite useful for very large packages. Changes are in commit c303564

[th0ma7]

@publicarray you might be interested by this as a starting point... Although road is quite long ahead.

My understanding is that (and feel free to let me know how off I may be):

1. using this approach we could, in theory, now create a C compiler using host=x86_64 and target=$(TC_TARGET)
2. using this new C compiler we should be able to create a full gcc suite to be used as a) target native package (e.g. enhance this package PR) and b) as host=x86_64 and target=$(TC_TARGET) set of compiler tools for subsequent steps.
3. using the full gcc suite in host=x86_64 and target=$(TC_TARGET) we should now be able to build a newer version of gcc. Again, we might have to iterate steps 2 and 3 a few times to increase gcc versions to meet the minimal requirements as we progress towards the latest available gcc.

[hgy59]

This is a candidate for another synocli package synocli-dev (or synocli-devel or synocli-develop) as proposed in #3253.

[hgy59]

I was working on synocli-develop some time ago.
The progess was until

cross/bc cross/jq cross/binutils cross/make cross/bison cross/flex

but then I struggled with cross/glibc.

It would be awsome to have develop tools on the target devices. For some extreme situations it would allow to create native binaries directly on the devices.

[th0ma7]

@hgy59 that really is an excellent idea...

To get glibc going you most probably need the gcc part first.

I was thinking about the feasibility of creating a new dev mk helper in order to build updated toolchain (host=x86_64 and target=$(TC_TARGET)). Then use that to build the updated synocli-dev environment for target only environment.

[publicarray]

Hi @th0ma7 thanks so much for your work on this. At the moment I'm focused on getting the .net packages (*arr) out the door, but will look at this after.

[th0ma7]

Hi @publicarray as an FYI I started preliminary work on recreating gcc cross-compiler. But I'm also working on getting the kernel module packages out of the door so hoping to resume work on this after.