Q: iPad Safari crashing regularly

JimHdk wrote:

 

Stephen Bennett1 wrote:

 

That simple? This issue has been around long enough for Apple to fix.

In complex multi-threaded software, bad pointer usage is one of the most difficult problems to find and fix. The problems are not reliably reproducible since they are affected by the order in which the end-user does things as well as being dependent on what he was doing previously (the state of the software).

 

The bad pointer could be the result of bad logic (release then reuse) or it could be do to a random store into the data area containing the pointer thus corrupting it. There are many other scenarios as well.

 

It's not "simple" at all.

I disagree. A principle software engineer working in the multi-thread space should, assuming the system is designed correctly and documented, be in a position to pinpoint such issues in a matter of days.

And actually in iOS 5.0.1 Apple has the new memory garbage collection which allows developers to not have to think about managing releasing objects.  Could be that it doesn't work perfectly yet, even for Safari.

Kinda think that might be related.

Bill

wdr99 wrote:

 

And actually in iOS 5.0.1 Apple has the new memory garbage collection which allows developers to not have to think about managing releasing objects.  Could be that it doesn't work perfectly yet, even for Safari.

 

Kinda think that might be related.

 

Bill

Interesting.  I haven't moved to iOS 5 yet.

I was thinking that too, but it looks like it ("Automatic Reference Counting") is a compiler feature rather than an OS/runtime feature, so it isn't actually garbage collection. In that case, wouldn't it depend on the individual app to be compiled with ARC or not?

Hmm... Without being actually being familiar with the specific code involved that's a rather extraordinary claim.

It's also possible that a problem has been identified but the solution and testing of same may take time given the variety of environments which must be tested. This can't be easy given that the original problem in iOS 5 escaped Apple's in-house and beta testing which is, presumably, quite extensive.

As I said before we can speculate as to cause (and we may be way off-base) but we must await a solution from Apple.

PS: I was actually at a higher level than Principal Software Engineer (note that's Principal not Principle) before I retired.

I only offer ARC as a fun theory due to my own development background and as an answer to the "What's changed?" question.  Pure speculation but it seems like everything in my crash log is bad address type errors.  Also, I'm definitely seeing less crashing in Safari with JavaScript turned off. 

Related or not, large iTunes pages seem to blow me up, too. 

Bill

Diavonex:

1. This does not work.

2. Here is an article showing why this should not work, nor need to be done.

http://speirs.org/blog/2012/1/2/misconceptions-about-ios-multitasking.html

(or google: fraser speirs ios multitasking)

I wish it did work, this would be a simple workaround.

I find that the most reliable way to crash the iPad1 is to go to Apple's own web site, click on the LION page, then try to load the 'explore new features in Lion'.  This crashes it every time on my iPad1.

JimHdk wrote:

 

Hmm... Without being actually being familiar with the specific code involved that's a rather extraordinary claim.

 

If you care to step outside your tardis you will find it is not extraordinary at all, if the system design is well thought out and documented. In software engineering today lingering issues tend to be associated with either design flaws or simple lack of knowledge.

Yes, in a perfect world all design is well thought out and perfect. However, the error in question could easily be a design error (e.g., an overlooked need for a lock, etc.). It's unlikely to be a simple coding error.

With the presumption of a well thought out design in place I guess the error couldn't have happened in the first place.

Perhaps the HAL 9000 could solve our problems. Oops, I guess it had a design error.

if the system design is well thought out and documented.

I've been an engineer in product development for nearly 25 years... I've never lived in, seen or heard of that world.

paulcb wrote:

 

if the system design is well thought out and documented.

I've been an engineer in product development for nearly 25 years... I've never lived in, seen or heard of that world.

I understand. Most software products don't comply to the standards and SLAs consumers expect.

JimHdk wrote:

 

Yes, in a perfect world all design is well thought out and perfect. However, the error in question could easily be a design error (e.g., an overlooked need for a lock, etc.). It's unlikely to be a simple coding error.

 

With the presumption of a well thought out design in place I guess the error couldn't have happened in the first place.

 

Perhaps the HAL 9000 could solve our problems. Oops, I guess it had a design error.

My point is not that error can't occur. It's that when error does occur you have the capacity to correct it efficiently.

Hey apple. Yeah it's still crashing. Please fix it. The Internet is kind of a big priority on the ipad

Use this

http://www.apple.com/feedback/ipad.html

How disappointing and frustrating!!!!

I have just received my Ipad2 about 4 hours ago, and Safari is already crashing.

What it really annoys me the most it is the fact that this problem has been going on for quite few time, and so far no solution to it.

Does anyone recommend any workaround to this problem? Maybe Opera or any other application can do the job as Safari is a nightmare?

Thanks in advance.