Re: Determining how the app is run (intel/ppc/rosetta/os version)?
Re: Determining how the app is run (intel/ppc/rosetta/os version)?
- Subject: Re: Determining how the app is run (intel/ppc/rosetta/os version)?
- From: Gerben Wierda <email@hidden>
- Date: Sun, 21 May 2006 12:15:04 +0200
On 21 May 2006, at 12:00, Finlay Dobbie wrote:
On 21/05/06, Gerben Wierda <email@hidden> wrote:
Because my app ran fine as UB in 10.4.3 on a DTK even though there
was an unexplained problem. This problem I could not solve (deep
inside Apple's Frameworks somehow an NSData object was produced with
a negative length, to put it shortly) but luckily it only gave a
warning in console.app when the NSData was changed in an NSString
(this was subprocess output data) and the NSString (nil) was used in
an NSView. That way, my app ran for years.
Ah yes, an intermittent bug that you ignored because it sort of worked
and then came back to bite you later. Are you entirely sure it's
Apple's fault? Did you complain to DTS about it?
Should we let the technical question slip away like this?
But, to answer your question: no of course I am not 100% certain I am
not doing anything wrong. This is why, when I was unable to find this
out myself and because the visible problem originated so deep in the
frameworks that I could not debug it I had to call on Apple with a
support ticket. This was february and Apple did not really do
anything until I got an apology lately that they had dropped the ball
on me and were now picking it up. So far, without success, but we're
fresh on our way so I'll wait (again).
So, yes, I am in contact with DTS about it and so far without
success. My did not 'sort of work', the code ran fine on 10.2, the
problems appeared with 10.3 without the code changed. On 10.4.3 it
still rans as on 10.3.9 and as of 10.4.4 the trouble started.
I am providing an open source support for free for (mainly) the TeX
community on intel. My time is very limited because I also have a
family and have to earn a living. So, yes, when I spent a lot of time
and was unable to find the source of the problem (most likely some
nasty combination of Security Framework, multithreading, distributed
objects, named pipes and whatever hoops I had to jump to get a decent
NSTask-like interface with AutorizationExecuteWithPrivileges). So as
with my many users the problem was as harmless as a warning and the
app was rock solid and always did what it should, I did choose not to
pursue it further. Now, the moment is there that I have to (hence the
DTS ticket without much success)
So, I now have to put out an app that does not run on 10.4.4 and up
on intel. And if I fix it such that it works via rosetta, I would
like to be able to let it run as a rosetta app but not native. I want
this check in my code so that it does not depend on my compile.
That's not really a proper fix, but it's certainly possible. See
<http://developer.apple.com/documentation/MacOSX/Conceptual/
universal_binary/universal_binary_exec_a/chapter_7_section_7.html#//
apple_ref/doc/uid/TP40002217-CH210-BAJEBJJF>.
That helps for rosetta. I'll look into sysctl once more. I'll already
use it to find out if a authorized subprocess is (still) running.
-- Finlay
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Cocoa-dev mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden