I made the mistake of installing XCode 14 today,
at Apple's prompting and it completely broke the
exception handling in variouscomponents of my KDF9
emulation software.

I was able to get back to the previous release of XCode
from a backup, and things seem to be OK again.

At any rate, the regression tests are not failing any more.

I am using FSF gnat-12.1.0-1, and it seems that there
is now a gnat-12.1.0-2 release, but I do not see
anything there that is relevant to my problem.

Any insight anyone?
Why would an updated XCode have such a drastic effect?

--
Bill Findlay

Bill Findlay <findlaybill@blueyonder.co.uk> writes:

> I made the mistake of installing XCode 14 today,
> at Apple's prompting and it completely broke the
> exception handling in variouscomponents of my KDF9
> emulation software.
>
> I was able to get back to the previous release of XCode
> from a backup, and things seem to be OK again.
>
> At any rate, the regression tests are not failing any more.
>
> I am using FSF gnat-12.1.0-1, and it seems that there
> is now a gnat-12.1.0-2 release, but I do not see
> anything there that is relevant to my problem.
>
> Any insight anyone?
> Why would an updated XCode have such a drastic effect?

If there's any C++ about, the troubles I had with GNT-LLVM[1] might
relate?

[1] https://forward-in-code.blogspot.com/2022/03/gnat-llvm.html

Bill Findlay <findlaybill@blueyonder.co.uk> writes:

> I made the mistake of installing XCode 14 today,
> at Apple's prompting and it completely broke the
> exception handling in variouscomponents of my KDF9
> emulation software.
>
> I was able to get back to the previous release of XCode
> from a backup, and things seem to be OK again.
>
> At any rate, the regression tests are not failing any more.
>
> I am using FSF gnat-12.1.0-1, and it seems that there
> is now a gnat-12.1.0-2 release, but I do not see
> anything there that is relevant to my problem.
>
> Any insight anyone?
> Why would an updated XCode have such a drastic effect?

Well, I just tried, and it was different for me: both on Intel and Apple
silicon any attempt to compile (or use git ...) failed dismally
somewhere in Xcode. On Apple silicon, I thought it was a failed update
so deleted Xcode and installed the command line tools, OK. Now, on
Intel, going to move Xcode out of the way ... and building with the
CLTs (version 14.0.0.0.1.1661618636) works fine.

On 16 Sep 2022, Simon Wright wrote
(in article <lybkrfmlmb.fsf@pushface.org>):

> Bill Findlay<findlaybill@blueyonder.co.uk> writes:
>
> > I made the mistake of installing XCode 14 today,
> > at Apple's prompting and it completely broke the
> > exception handling in variouscomponents of my KDF9
> > emulation software.
> >
> > I was able to get back to the previous release of XCode
> > from a backup, and things seem to be OK again.
> >
> > At any rate, the regression tests are not failing any more.
> >
> > I am using FSF gnat-12.1.0-1, and it seems that there
> > is now a gnat-12.1.0-2 release, but I do not see
> > anything there that is relevant to my problem.
> >
> > Any insight anyone?
> > Why would an updated XCode have such a drastic effect?
>
> If there's any C++ about, the troubles I had with GNT-LLVM[1] might
> relate?
>
> [1] https://forward-in-code.blogspot.com/2022/03/gnat-llvm.html

C++ does not darken my doors. 8-)

--
Bill Findlay

On 16 Sep 2022, Simon Wright wrote
(in article <ly7d23ml26.fsf@pushface.org>):

> Bill Findlay<findlaybill@blueyonder.co.uk> writes:
>
> > I made the mistake of installing XCode 14 today,
> > at Apple's prompting and it completely broke the
> > exception handling in variouscomponents of my KDF9
> > emulation software.
> >
> > I was able to get back to the previous release of XCode
> > from a backup, and things seem to be OK again.
> >
> > At any rate, the regression tests are not failing any more.
> >
> > I am using FSF gnat-12.1.0-1, and it seems that there
> > is now a gnat-12.1.0-2 release, but I do not see
> > anything there that is relevant to my problem.
> >
> > Any insight anyone?
> > Why would an updated XCode have such a drastic effect?
>
> Well, I just tried, and it was different for me: both on Intel and Apple
> silicon any attempt to compile (or use git ...) failed dismally
> somewhere in Xcode. On Apple silicon, I thought it was a failed update
> so deleted Xcode and installed the command line tools, OK. Now, on
> Intel, going to move Xcode out of the way ... and building with the
> CLTs (version 14.0.0.0.1.1661618636) works fine.

Thanks for getting on this, Simon.

I have now deleted even the older XCode and things continue OK,
now withgnat-12.1.0-2.

How can I find out the version number of the CLTs?

--
Bill Findlay

On 16 Sep 2022, Bill Findlay wrote
(in article<0001HW.28D4D4170006908270000F0D638F@news.individual.net>):

> On 16 Sep 2022, Simon Wright wrote
> (in article <ly7d23ml26.fsf@pushface.org>):
>
> > Bill Findlay<findlaybill@blueyonder.co.uk> writes:
> >
> > > I made the mistake of installing XCode 14 today,
> > > at Apple's prompting and it completely broke the
> > > exception handling in variouscomponents of my KDF9
> > > emulation software.
> > >
> > > I was able to get back to the previous release of XCode
> > > from a backup, and things seem to be OK again.
> > >
> > > At any rate, the regression tests are not failing any more.
> > >
> > > I am using FSF gnat-12.1.0-1, and it seems that there
> > > is now a gnat-12.1.0-2 release, but I do not see
> > > anything there that is relevant to my problem.
> > >
> > > Any insight anyone?
> > > Why would an updated XCode have such a drastic effect?
> >
> > Well, I just tried, and it was different for me: both on Intel and Apple
> > silicon any attempt to compile (or use git ...) failed dismally
> > somewhere in Xcode. On Apple silicon, I thought it was a failed update
> > so deleted Xcode and installed the command line tools, OK. Now, on
> > Intel, going to move Xcode out of the way ... and building with the
> > CLTs (version 14.0.0.0.1.1661618636) works fine.
>
> Thanks for getting on this, Simon.
>
> I have now deleted even the older XCode and things continue OK,
> now with gnat-12.1.0-2.
>
> How can I find out the version number of the CLTs?

Well, I spoke too soon.
Sometime between last night and this morning it reverted to non-working!

Same problem: compiled programs collapse when an exception is raised.

I have deleted and re-installed the CLTs using xcode-select, but no joy.
Snookered!

--
Bill Findlay

Bill Findlay <findlaybill@blueyonder.co.uk> writes:

> How can I find out the version number of the CLTs?

$ pkgutil --pkg-info=com.apple.pkg.CLTools_Executables

(on older OSs there are different forms)

On 16 Sep 2022, Simon Wright wrote
(in article <ly35crm99x.fsf@pushface.org>):

> pkgutil --pkg-info=com.apple.pkg.CLTools_Executables

Thanks, version 14.0.0.0.1.1661618636 is what I have.

I got back to sanity by restoring
/Library/Developer/CommandLineTools/usr/bin
from a backup.

The (working) tools in there are over a year old.
I wonder what Apple did to bork them with XCode 14.

--
Bill Findlay

Bill Findlay <findlaybill@blueyonder.co.uk> writes:

> Well, I spoke too soon.
> Sometime between last night and this morning it reverted to non-working!
>
> Same problem: compiled programs collapse when an exception is raised.
>
> I have deleted and re-installed the CLTs using xcode-select, but no joy.
> Snookered!

Just re-ran ACATS 4.1Y - abandoned after this:

Running chapter c2 ...
FAIL: c250002
Running chapter c3 ...
FAIL: c324006
FAIL: c34007j
FAIL: c380003
FAIL: c390002
Running chapter c4 ...
FAIL: c415001
FAIL: c43206a
FAIL: c45304a
FAIL: c45504a
FAIL: c45632a
FAIL: c46014a
FAIL: c48006b
Running chapter c5 ...

and this in the log

RUN c43206a

,.,. C43206A ACATS 4.1Y 22-09-16 20:29:49
---- C43206A CHECK THAT THE BOUNDS OF A NULL ARRAY ARE DETERMINED BY THE
BOUNDS SPECIFIED BY THE CHOICES.
libunwind: _Unwind_GetDataRelBase - _Unwind_GetDataRelBase() not implemented
libunwind: _Unwind_GetDataRelBase - _Unwind_GetDataRelBase() not implemented
libunwind: _Unwind_GetDataRelBase - _Unwind_GetDataRelBase() not implemented
libunwind: _Unwind_GetDataRelBase - _Unwind_GetDataRelBase() not implemented
libunwind: _Unwind_GetDataRelBase - _Unwind_GetDataRelBase() not implemented
libunwind: _Unwind_GetDataRelBase - _Unwind_GetDataRelBase() not implemented
...

so it seems there is an unwinding problem.

Simon Wright <simon@pushface.org> writes:

> Well, I just tried, and it was different for me: both on Intel and
> Apple silicon any attempt to compile (or use git ...) failed dismally
> somewhere in Xcode.

Turns out you have to actually *run* the new Xcode (and make the default
choice of components) for it to install its version of the CLTs,
including git.

On 22 Sep 2022, Simon Wright wrote
(in article <lybkr7svre.fsf@pushface.org>):

> Simon Wright <simon@pushface.org> writes:
>
> > Well, I just tried, and it was different for me: both on Intel and
> > Apple silicon any attempt to compile (or use git ...) failed dismally
> > somewhere in Xcode.
>
> Turns out you have to actually *run* the new Xcode (and make the default
> choice of components) for it to install its version of the CLTs,
> including git.

So, having run XCode, do you get the exception failure with the new CLTs?

--
Bill Findlay

Bill Findlay <findlaybill@blueyonder.co.uk> writes:

> On 22 Sep 2022, Simon Wright wrote
> (in article <lybkr7svre.fsf@pushface.org>):
>
>> Simon Wright <simon@pushface.org> writes:
>>
>> > Well, I just tried, and it was different for me: both on Intel and
>> > Apple silicon any attempt to compile (or use git ...) failed dismally
>> > somewhere in Xcode.
>>
>> Turns out you have to actually *run* the new Xcode (and make the default
>> choice of components) for it to install its version of the CLTs,
>> including git.
>
> So, having run XCode, do you get the exception failure with the new CLTs?

Short answer: yes, fraid so.

Looks as though 14.1 is going to have the same problem (given that the
beta has). But we might be lucky.

(/Library/Developer/CommandLineTools contains a subset of the equivalent
/Applications/Xcode.app)

On 22 Sep 2022, Simon Wright wrote
(in article <ly5yhf18ln.fsf@pushface.org>):

> Bill Findlay<findlaybill@blueyonder.co.uk> writes:
>
> > On 22 Sep 2022, Simon Wright wrote
> > (in article <lybkr7svre.fsf@pushface.org>):
> >
> > > Simon Wright <simon@pushface.org> writes:
> > >
> > > > Well, I just tried, and it was different for me: both on Intel and
> > > > Apple silicon any attempt to compile (or use git ...) failed dismally
> > > > somewhere in Xcode.
> > >
> > > Turns out you have to actually *run* the new Xcode (and make the default
> > > choice of components) for it to install its version of the CLTs,
> > > including git.
> >
> > So, having run XCode, do you get the exception failure with the new CLTs?
>
> Short answer: yes, fraid so.
>
> Looks as though 14.1 is going to have the same problem (given that the
> beta has). But we might be lucky.
>
> (/Library/Developer/CommandLineTools contains a subset of the equivalent
> /Applications/Xcode.app)

Thanks for the info, Simon.

--
Bill Findlay