Meego vs Android apps (C or Java?) - Page 16 - maemo.org

Active Topics

The N900 is now 15 years old! (9)
to Nokia N900 by Arno_11 - 22 hrs, 5 mins ago
Sailfish OS for the Motorola Moto G7 Power (XT1955-5) - (ocean) (4)
to SailfishOS by Maemish - 6 days, 10 hrs ago
more...

Page 16 of 21

Thread Tools

	nwerneck	2011-02-10 , 21:54
	Posts: 304 \| Thanked: 233 times \| Joined on Jul 2009 @ São Paulo, SP, Brasil	#151

Originally Posted by attila77

As I said previously, a lot of Java people think about C++ as this awkward C bolt-on from the '80s, and C++ people think of Java as this academic bloatware. Well, dears coders and codettes, it's 2011 and neither is true.

Totally agreed. I use C myself, and don't know much about Java, but what I really like is Python! =) And I must say I code in Python always very much aware of where all inner pointers and gotos are pointing at. I try to keep in mind that programming languages are illusions over the real bit flipping operations, and all these seemingly important debates lose importance to me.

But I am really replying just to share this quote here with you guys:

"C++ is history repeated as tragedy. Java is history repeated as farce." --- Scott McKay

Quote & Reply |

The Following User Says Thank You to nwerneck For This Useful Post:
MartinK

Joorin	2011-02-10 , 22:11
Posts: 726 \| Thanked: 345 times \| Joined on Apr 2010 @ Sweden	#152

Originally Posted by attila77

This is the out-of-thin-air generalization that makes makes even your good points look bad. D has pointers. C# has pointers. Hell, even Google's brand spanking new Go language has pointers.

Just to add a tiny nuance: C# offers pointers if you agree to be on your own, taking full responsibility for it all.

And when you do, optimizing the code gets harder (for the more interesting cases than just slipping a pointer to a native library call).

If you stay away from such endeavours, C# is not offering any pointers for the programmer to use.

With the usual AFAIK disclaimer.

Quote & Reply |

The Following User Says Thank You to Joorin For This Useful Post:
zimon

v13	2011-02-11 , 09:45
Posts: 376 \| Thanked: 511 times \| Joined on Aug 2009 @ Greece	#153

Originally Posted by zimon

The problem is obvious for me, but I know I should prove it with memory prints comparing the same session and tabs and pages after fresh start and after few days of surfing. Also I could show how cache misses will increase over time with Firefox. I've actually sometimes done these observations but not in a scientific way yet.

I really can't blindly accept what the post says. A long time ago, in a version far far away, glibc started using mmap() on /dev/zero instead of brk() to allocate memory. This has the advantage that memory that is freed returns to the OS immediately and Linux is able to handle memory fragmentation better than Firefox itself (memory fragmentation on OS level is a very different thing). This means that there needs to be no free memory handling in any application, unless the app chooses to re-use its freed memory. But even in that case, fragmentation could be reduced by just freeing fragmented memory.

So, no. When judging languages leave the memory fragmentation out of the conversation and surely don't compare the memory handling of Java with the memory handling of Firefox. I'm 100% sure that the Llinux kernel can do much better than JVM since it has actual view of the physical ram and the actual page allocation.

Quote & Reply |

zimon	2011-02-11 , 16:35
Posts: 1,341 \| Thanked: 708 times \| Joined on Feb 2010	#154

This whole thread is moot now.
Should you change topic to "WP7 vs Android apps (C# or Java?)"

When I have time, I try to prove that slowing down of my Firefox is due to heap memory fragmentation. No *malloc can remove fragmentation totally, because alive objects cannot be moved in the heap if pointers are allowed in the code (C/C++). I could point to some studies done for GC+memory management, but you guys can find them yourselves also with google.

And yes, Java also kind of still have pointers, via JNI kinda, but as with C#, optimizations go down the drain then on that basic block of the code where the pointers are used.

Last edited by zimon; 2011-02-11 at 19:48. Reason: smile removed. it is a sad day. 11.2. as 112

Quote & Reply |

attila77	2011-02-12 , 18:36
Posts: 3,319 \| Thanked: 5,610 times \| Joined on Aug 2008 @ Finland	#155

Originally Posted by Joorin

Just to add a tiny nuance: C# offers pointers if you agree to be on your own, taking full responsibility for it all.

In what way is that different from C++ ?

Admittedly the 'usual C++ way' is heavy with pointer use, but that's not a requirement by any means.

Originally Posted by zimon

When I have time, I try to prove that slowing down of my Firefox is due to heap memory fragmentation.

Just make sure Flash and javascript are disabled as those might prevent the release of memory.

No *malloc can remove fragmentation totally, because alive objects cannot be moved in the heap if pointers are allowed in the code (C/C++).

That is sort-of true for *unmanaged* pointers, but for example if you use Qt, most of your objects would be QObjects or wrapped in smart pointer classes which allow Qt to do whatever it wants memory-wise as it 'knows' about what is where.

I could point to some studies done for GC+memory management, but you guys can find them yourselves also with google.

This is a fairly difficult question for several reasons. One reason that you do not know the link between the heap and the physical memory (swap, virtualization, etc). Thus, if suddenly your memory manager decides 'hey if I move this there might be space for that new object' it might actually cause a performance decrease depending on what needs to be done in heapspace. If a VM tries to take over the OS's memory management function, that's always a dubious outcome.

__________________
Blogging about mobile linux - The Penguin Moves!
Maintainer of PyQt (see introduction and docs), AppWatch, QuickBrownFox, etc

Quote & Reply |

Joorin	2011-02-12 , 18:52
Posts: 726 \| Thanked: 345 times \| Joined on Apr 2010 @ Sweden	#156

Originally Posted by attila77

In what way is that different from C++ ? Admittedly the 'usual C++ way' is heavy with pointer use, but that's not a requirement by any means.

C++

Code:

 Foo* foo = new Foo();

Code:

unsafe {
  Foo *foo = &someFoo;
}

The interesting difference, in this context, is that in C#, you must to be in an unsafe context/scope to be able to use pointers. In C++ it's just another construct, needing no special handling to be used.

Quote & Reply |

The Following User Says Thank You to Joorin For This Useful Post:
attila77

attila77	2011-02-12 , 21:58
Posts: 3,319 \| Thanked: 5,610 times \| Joined on Aug 2008 @ Finland	#157

Yes, I understand that, but IMO that's more syntactic sugar/policy question than a technical differentiation. The choice is still yours, it's just that you have less 'are you sure' dialogs

__________________
Blogging about mobile linux - The Penguin Moves!
Maintainer of PyQt (see introduction and docs), AppWatch, QuickBrownFox, etc

Quote & Reply |

Joorin	2011-02-12 , 23:57
Posts: 726 \| Thanked: 345 times \| Joined on Apr 2010 @ Sweden	#158

Originally Posted by attila77

Yes, I understand that, but IMO that's more syntactic sugar/policy question than a technical differentiation.

Huh? If there is something it isn't, it's syntactical sugar.

By declaring an unsafe context/scope you bypass the strict typing checks (not totally, but almost) and if you use fixed pointers you bypass the GC, not letting it move the pointed at memory.

This isn't sugar in any way, shape or form. This is affecting the compiler and the runtime behaviour.

The choice is still yours, it's just that you have less 'are you sure' dialogs

Sure, you can stay away from it, but if you use it, you need to use these special constructs, bypassing things and affecting others. Which was my point.

C# offers pointers, but not in the same way as C++, or C. If you use them, you're letting go of some fundamental features of the language along the way.

Quote & Reply |

The Following User Says Thank You to Joorin For This Useful Post:
attila77

	Capt'n Corrupt	2011-02-27 , 13:32
	Posts: 3,524 \| Thanked: 2,958 times \| Joined on Oct 2007 @ Delta Quadrant	#159

Sorry to drag this conversation from the depths, but..

Wouldn't it be possible to create a object/class that emulates pointer memory modification in a safe way?

It could potentially use the array primitive as the heap object to avoid the costly method call per pointer op and provide the mechanisms to do basic [pointer] arithmetic. It could also contain handy functionality to automatically re-size the object as memory needs increase/decrease to more closely emulate C's malloc.

Since all objects are passed as references in method calls, this could give different aspects of an application access to the same structures in a safe way -- and very little computational penalty. Also data structures would be totally isolated from one another, which is a definite plus.

While I appreciate that for a great many cases, objects and references are a fine substitute for the pointer, I also recognize that more exotic structures can use pointers to great effect. Perhaps this is a way to do it in the modern VM space.

Quote & Reply |

The Following User Says Thank You to Capt'n Corrupt For This Useful Post:
attila77

attila77	2011-02-27 , 20:03
Posts: 3,319 \| Thanked: 5,610 times \| Joined on Aug 2008 @ Finland	#160

Originally Posted by Joorin

Huh? If there is something it isn't, it's syntactical sugar.

By declaring an unsafe context/scope you bypass the strict typing checks (not totally, but almost) and if you use fixed pointers you bypass the GC, not letting it move the pointed at memory.

I was referring to it more in a policy context - a 'friendly' compiler could generate warnings on such code which you could 'declare' right by definint adequate pragmas (=you can achieve the same effect). Yes, syntactic sugar is not the most matching term, but I know no better how to describe this 'more than one way to do it' thing

Originally Posted by Capt'n Corrupt

Wouldn't it be possible to create a object/class that emulates pointer memory modification in a safe way?

It could potentially use the array primitive as the heap object to avoid the costly method call per pointer op and provide the mechanisms to do basic [pointer] arithmetic. It could also contain handy functionality to automatically re-size the object as memory needs increase/decrease to more closely emulate C's malloc.

QPointer and friends *sort of* do this. My code is mostly devoid of class* thingies because it's easier and safer to use Qt's smart pointer classes. While this does not do object movement in memory (*now* - as said above, benefits of that can be ambiguous), it does affect memory management, as it provides reference counting, CoW operations, shared data, etc.

Which brings things us to the (repeated) bottom line - C++ as such does not *require* a certain type of memory management, it's not part of the language, so if you have a fancier, smarter way of dealing with allocated memory, give it a go.

__________________
Blogging about mobile linux - The Penguin Moves!
Maintainer of PyQt (see introduction and docs), AppWatch, QuickBrownFox, etc