[MacPorts] #65147: nettle fails to build for x86_64 with gcc10-bootstrap: sha256-compress.asm:208:no such instruction

MacPorts noreply at macports.org
Tue May 10 01:59:12 UTC 2022


#65147: nettle fails to build for x86_64 with gcc10-bootstrap:
sha256-compress.asm:208:no such instruction
---------------------------+---------------------------------
  Reporter:  barracuda156  |      Owner:  ryandesign
      Type:  defect        |     Status:  assigned
  Priority:  Normal        |  Milestone:
 Component:  ports         |    Version:  2.7.2
Resolution:                |   Keywords:  x86_64, snowleopard
      Port:  nettle        |
---------------------------+---------------------------------

Comment (by barracuda156):

 Let me explain in some detail, so that I don't make an impression of doing
 something ridiculous and creating problems out of nowhere.

 As you may remember, I was working on getting `ghc` built for PPC. Long
 story short, `ghc` uses itself to build, so I need a working `ghc` for
 Intel to begin with. I went to Haskell site, and well, there is no support
 for 10.6 Intel (I did try to build what they had for 10.7, it failed, both
 with Xcode 3.2.6 and Xcode 4.2). I don't need it on 10.7, that gonna
 create a whole range of unneeded complications later for cross-building.
 So I went with Macports route instead.
 `ghc` in Macports needs a modern compiler, and I cannot use `gcc11` which
 I already built for `x86_64` before – outside of Macports – for above-
 mentioned purpose. The optimal way to get ''some'' compiler supporting
 2011 standard is `gcc10-bootstrap`. So I built that (tried universal, it
 failed, rebuilt again just for `x86_64`).
 Then `ghc` has a list of dependencies, which in turn require few ports
 that need 2011 standard compiler. So I wanted to save some effort and
 avoid building `gcc11` again, twice (because of `libgcc11`). I already
 have built `gcc` for `x86_64` 4 times just to get this thing going. And
 the only point is to build `ghc`, otherwise I have no use case for these.

 So while my case is rather uncommon, I don't think it is something crazy
 or absolutely unique. That was simply the fastest way to build `ghc` for
 Intel, given 10.6.8.

-- 
Ticket URL: <https://trac.macports.org/ticket/65147#comment:8>
MacPorts <https://www.macports.org/>
Ports system for macOS


More information about the macports-tickets mailing list