[PD] Attaching Xcode debugger
Pierre Alexandre Tremblay
tremblap at gmail.com
Tue Feb 28 12:54:27 CET 2023
Thanks for these
I was unclear - I’m trying to debug an external (all flags ok) to a downloaded version of Pd and this is where the sandboxing is freaking out lldb. In the article I pointed, they show how to modify the entitlements of the host, which I did successfully for a downloaded version of SuperCollider, but I don’t seem able to do this with the downloaded version of Pd...
> On 28 Feb 2023, at 11:51, Dan Wilcox <danomatika at gmail.com> wrote:
>
> Ah again... if you are debugging an external, it should also be compiled with optimization disabled and debugging symbols enabled, typically CFLAGS="-O0 -g"
>
>> On Feb 28, 2023, at 12:43 PM, Dan Wilcox <danomatika at gmail.com <mailto:danomatika at gmail.com>> wrote:
>>
>> Ah, I forgot an important point: first configure with debugging symbols.
>>
>> ./configure --enable-debug
>>
>> If building from a git clone, first generate configure using autoconf via:
>>
>> ./autogen.sh
>>
>> Actually, this info is in the INSTALL.txt file, but I'm writing it here anyway. ;)
>>
>>> On Feb 28, 2023, at 12:39 PM, Dan Wilcox <danomatika at gmail.com <mailto:danomatika at gmail.com>> wrote:
>>>
>>>
>>>> On Feb 28, 2023, at 12:00 PM, pd-list-request at lists.iem.at <mailto:pd-list-request at lists.iem.at> wrote:
>>>>
>>>> Hello
>>>>
>>>> So we all know how Apple is sandboxing like hell.
>>>
>>> Sure, but allowing *any* old program to be debugged / decompiled in realtime is not the best for security implications, at least Apple thinks so.
>>>
>>> Your subject is a little misleading as the "Xcode debugger" is essential lldb. Xcode itself simply wraps lldb when debugging and provides a UI top interact with it. In fact, on a debug build crash in Xcode, you can enter lldb commands directly into the console, ie. "bt" will provide a backtrace directly.
>>>
>>>> How do people attach a debugger to pd to do development?
>>>
>>> "lldb" like "gdb"
>>>
>>> For quick tests of a known crasher, I usually use:
>>>
>>> make
>>> cd bin
>>> lldb pd
>>>
>>> This will use the system Tcl/Tk which has had issues for years now, but is enough for quick testing.
>>>
>>> For best testing, build the .app and invoke the pd binary inside it. For example:
>>>
>>> make
>>> make app
>>> lldb Pd-0.53-2.app/Contents/Resources/bin/pd
>>>
>>> Once lldb launches, you have to start the process:
>>>
>>> process launch
>>>
>>> When a crash happens, you can get the backtrace with "bt", change frames with "f #", and print variables via "v". If you have used gdb before, this is a helpful crib sheet:
>>>
>>> https://lldb.llvm.org/use/map.html <https://lldb.llvm.org/use/map.html>
>>>
>>> Note: If you are using a newer macOS system and want to *also* use a newer version of Tcl/Tk, build Wish.app first then use it to make the Pd .app bundle:
>>>
>>> ./macos/tcltk-wish.sh 8.6.12
>>>
>>> produces a Wish-8.6.12.app. Next use it to build Pd .app with:
>>>
>>> ./mac/osx-app.sh -w Wish-8.6.12.app 0.53-2
>>>
>>> Now you can change the C sources and rebuild with:
>>>
>>> make
>>> ./mac/osx-app.sh -w Wish-8.6.12.app 0.53-2
>>>
>>>> Do I have to recompile pd myself as the long winded detour?
>>>
>>> If you want debugging symbols, yes. Fortunately it is relative easy to compile Pd so I don't consider it "long-winded." In fact, we have spent quite a bit of time to make this both possible *and* relatively easy. (See above.)
>>>
>>> --------
>>> Dan Wilcox
>>> @danomatika <http://twitter.com/danomatika>
>>> danomatika.com <http://danomatika.com/>
>>> robotcowboy.com <http://robotcowboy.com/>
>>>
>>>
>>>
>>
>> --------
>> Dan Wilcox
>> @danomatika <http://twitter.com/danomatika>
>> danomatika.com <http://danomatika.com/>
>> robotcowboy.com <http://robotcowboy.com/>
>>
>>
>>
>
> --------
> Dan Wilcox
> @danomatika <http://twitter.com/danomatika>
> danomatika.com <http://danomatika.com/>
> robotcowboy.com <http://robotcowboy.com/>
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.puredata.info/pipermail/pd-list/attachments/20230228/20739cac/attachment-0001.htm>
More information about the Pd-list
mailing list