[tlu] gpg problem

[Herbert Schulz]

> On Feb 27, 2021, at 11:10 AM, Herbert Schulz <herbs at wideopenwest.com> wrote:
> 
> 
> 
>> On Feb 27, 2021, at 10:39 AM, Adam R. Maxwell via tlu <tlu at tug.org> wrote:
>> 
>> 
>> 
>>> On Feb 27, 2021, at 08:08 , Herbert Schulz <herbs at wideopenwest.com> wrote:
>>> 
>>> 2021-02-27 16:05:50 +0000 Notice -[TLMOperation main][70789]	start load http://www.preining.info/tlgpg/
>>> 
>>> /Library/TeX/texbin/tlmgr: signature verification error of /usr/local/texlive/2021/tlpkg/texlive.tlpdb.main.e43e83a9b003c5f44fae318dbaa6fde3 from http://www.preining.info/tlgpg//tlpkg/texlive.tlpdb: cryptographic signature verification of
>>> /var/folders/wh/5072hqh48xj1mw001s6v2kdh0000gn/T/XMoFDOS7Rk/fF_gPY556C
>>> against
>>> http://www.preining.info/tlgpg//tlpkg/texlive.tlpdb.sha512.asc
>>> failed. Output was:
>>> sh: /usr/local/texlive/2021/tlpkg/installer/gpg/gpg.universal-darwin: Bad CPU type in executable
>> 
>> No idea what's going on here, but my initial guess is that it's related to a new architecture for the Mac. Are you using an Intel or M1? I'm copying Norbert, so we can figure out how to debug it.
>> 
>> thanks,
>> Adam
> 
> Howdy,
> 
> I'm still on Intel (for the next few years!).
> 
> Well... I just ran
> 
> file /usr/local/texlive/2021/tlpkg/installer/gpg/gpg.universal-darwin
> 
> and got
> 
> /usr/local/texlive/2021/tlpkg/installer/gpg/gpg.universal-darwin: Mach-O universal binary with 2 architectures: [i386:Mach-O executable i386] [ppc_7400:Mach-O executable ppc_7400]
> /usr/local/texlive/2021/tlpkg/installer/gpg/gpg.universal-darwin (for architecture i386):	Mach-O executable i386
> /usr/local/texlive/2021/tlpkg/installer/gpg/gpg.universal-darwin (for architecture ppc7400):	Mach-O executable ppc_7400
> 
> So there is definitely a problem with gpg.universal-darwin. I suspect this is the ancient universal-darwin that was PPC/Intel32. I just knew there would be a problem there.
> 
> Good Luck,
> 
> Herb Schulz
> herbs at wideopenwest.com

Howdy,

I just copied the gpg.x86_64-darwin from TL2020 to gpg.universal-darwin in TL2021 and the problem goes away. It should run fine under Rosetta on M1 systems but someone should test it.

PS: I copied this to Dick since I know he has an M1 system.

Good Luck,

Herb Schulz
herbs at wideopenwest.com