[m-users.] Strange build problem with Mercury after installing Pandoc

[Sean Charles (emacstheviking)]

SOLVED! Why it is using x86_64 not arm64

I had, for the first time since using my M1, started the day using cool-retro-term, having just installed it...., during the course of diving into my problem, I did this in the console:

    $ arch
    i386

Whaaaaaaaaaat?

Turns out that the brew install is running an intel build using Rosetta and this is of course picked up by the build system, excitedly I ran iTerm2 (M1 version)...

    $ arch
    arm64

Phew! And yes my code now builds again.... so the lesson learned is make sure you aren't running Rosetta code when using compiler tools, unless you --want-- to be in i86 mode I guess!

Thanks all.




> On 18 Jun 2022, at 12:44, Sean Charles (emacstheviking) <objitsu at gmail.com> wrote:
> 
> Rebuilding from scratch is exactly how the problem manifested itself in the first place.
> The distribution of mercury on my machine was built from source on M1, without issue.
> 
> Mercury Compiler, version 22.01.1, on aarch64-apple-darwin21.4.0
> Copyright (C) 1993-2012 The University of Melbourne
> Copyright (C) 2013-2022 The Mercury team
> 
> 
> I am not sure why you suggest to install Pandoc using intel architecture as I am using an M1 ARM processor, Pandoc installed fine but since then Mercury will not link.
> 
> It's almost as though some setting somewhere has changed such that the linker thinks the default machine architecture is intel not arm64. 
> 
> Very annoying, a complete blocker for me, but as always with times like this... a learning opportunity!
> 
> Thanks
> Sean.
> 
> 
>> On 18 Jun 2022, at 10:06, Volker Wysk <post at volker-wysk.de <mailto:post at volker-wysk.de>> wrote:
>> 
>> Am Samstag, dem 18.06.2022 um 08:43 +0100 schrieb Sean Charles (emacstheviking):
>>> Hi,
>>> 
>>> I wondered if anybody had seen this and knew how to deal with it... for weeks I've been happily developing my program with Mercury on my new M1 macMini, but this morning I needed to install Pandoc using brew, but in order for it to work I had to use this command line, found from googling it:
>>> 
>>>     $ arch -arm64 brew install pandoc
>>> 
>>> I tested pandoc with a small markdown to pdf run and it was fine, then I attempted to rebuild my application and for the first time ever I got this error:
>>> 
>>> ld: warning: ignoring file /usr/local/mercury-22.01.1/lib/mercury/lib/libgc.dylib, building for macOS-x86_64 but attempting to link
>>>  with file built for macOS-arm64
>>> ld: warning: ignoring file /usr/local/mercury-22.01.1/lib/mercury/lib/hlc.gc/libmer_std.dylib, building for macOS-x86_64 but attemp
>>> ting to link with file built for macOS-arm64
>>> ld: warning: ignoring file /usr/local/mercury-22.01.1/lib/mercury/lib/hlc.gc/libmer_rt.dylib, building for macOS-x86_64 but attempt
>>> ing to link with file built for macOS-arm64
>>> Undefined symbols for architecture x86_64:
>>>   "_GC_init", referenced from:
>>>       _init_gc in mlpt_init.o
>>>   "_GC_malloc", referenced from:
>>>       <predicate 'classifier.classify_line'/3 mode 0> in classifier.o
>>>       <predicate 'classifier.classify_file'/4 mode 0> in classifier.o
>>>       _cleaner__markdown_out_3_p_0_1 in cleaner.o
>>>       <predicate 'splitter.lines_to_blocks'/2 mode 0> in splitter.o
>>>       _splitter__lines_to_blocks_2_p_0_1 in splitter.o
>>>       <predicate 'weaver.UnusedArgs__pred__code_out__[1]_0'/4 mode 0> in weaver.o
>>>       _weaver__duplicate_check_2_p_0_1 in weaver.o
>>>       ...
>>> 
>>> The full file is very very long but it seems to somehow think that the default build architecture is intel not arm64. I have tried googling but so far to no avail, it has affected all of my Mercury projects on my machine, the same issue during linking.
>> 
>> This may be a stupid question, but have you tried to rebuild your application from scratch? Something like "make clean; make depend" and then make the application.
>> 
>> Or, maybe, try to install pandoc with "arch -x86_64 brew install pandoc". (That's just guesswork by me.)
>> 
>> Good luck,
>> Volker
> 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mercurylang.org/archives/users/attachments/20220618/86ec9ea9/attachment.html>