[mono][ios] Add support for stripping during AOT compilation on iOS

[kotlarmilos]

This PR adds support for stripping debug symbols and enabling IL stripping during AOT compilation of the HelloiOS sample app.

• Added StripSymbolTable="$(StripDebugSymbols)" to the AppleBuild.targets file.
• Added <ShouldILStrip Condition="'$(RunAOTCompilation)' == 'true' and '$(MonoForceInterpreter)' != 'true'">true</ShouldILStrip> to the Program.csproj file.

This PR should resolve the reported SOD regression.

[ghost]

Tagging subscribers to this area: @directhex
See info in area-owners.md if you want to be subscribed.

Issue Details

---

This PR adds support for stripping debug symbols and enabling IL stripping during AOT compilation of the HelloiOS sample app.

• Added StripSymbolTable="$(StripDebugSymbols)" to the AppleBuild.targets file.
• Added <ShouldILStrip Condition="'$(RunAOTCompilation)' == 'true' and '$(MonoForceInterpreter)' != 'true'">true</ShouldILStrip> to the Program.csproj file.

This PR should resolve the reported SOD regression.

Author:	kotlarmilos
Assignees:	kotlarmilos
Labels:	area-Infrastructure-mono, os-ios
Milestone:	8.0.0

[kotlarmilos]

@steveisok, I noticed the commented ShouldILStrip property in the AppleBuild.props file. Is there an issue on the CI or can it be enabled for the library tests?

[steveisok]

@steveisok, I noticed the commented ShouldILStrip property in the AppleBuild.props file. Is there an issue on the CI or can it be enabled for the library tests?

I think there was a problem running it on libraries, but does not appear an issue was logged. Try to enable on libraries and see what we get.

[ivanpovazan]

LGTM!

Nit:
One small thing worth considering is adding in some top level import (like AppleBuild.props) a default value for StripDebugSymbols like:
<StripDebugSymbols Condition="'$(StripDebugSymbols)' == ''" >false</StripDebugSymbols>
I didn't find it being set anywhere except in the Makefile of the project.

[ivanpovazan]

A slightly off-topic comment, but related to HelloiOS SOD regressions:

There was also a regression reported with NativeAOT.
The regression seems to be introduced with: 2835110

However, the regression is justified, as for NativeAOT builds we were using feature switch <InvariantGlobalization>true</InvariantGlobalization> when building the sample project because the support for System.Globalization.Native was not implemented yet. This made the app smaller, but at the same time was not matching how Mono sample was configured.

This has been fixed in the mentioned change: 2835110#diff-41016a8344332e64360007e06a117aa072eac2efe3e0e0109ce27d5466695dccL17-L18
which makes the SOD of the NativeAOT app a bit bigger, but is now aligned with the Mono configuration, making the comparison fairer.

/cc: @SamMonoRT

[kotlarmilos]

/azp run runtime-ioslike,runtime-ioslikesimulator

[azure-pipelines]

Azure Pipelines successfully started running 2 pipeline(s).

[kotlarmilos]

/azp run runtime-ioslike,runtime-ioslikesimulator

[azure-pipelines]

Azure Pipelines successfully started running 2 pipeline(s).

[kotlarmilos]

Some of the failures are reported in #73040 and shouldn't be related.

[akoeplinger]

@steveisok, I noticed the commented ShouldILStrip property in the AppleBuild.props file. Is there an issue on the CI or can it be enabled for the library tests?

I think there was a problem running it on libraries, but does not appear an issue was logged. Try to enable on libraries and see what we get.

Yeah we had problems with ActiveIssue attributes not being retained so we disabled ILStrip a while ago in #59762

I think the underlying issue could have been fixed by #87923