www.digitalmars.com         C & C++   DMDScript  

digitalmars.D.learn - Question about garbage collector

reply "bioinfornatics" <bioinfornatics fedoraproject.org> writes:
Hi everyone,

yesterday i read an article into a french linux journal that in 
some years garbage collector will disapear.

Why ? he explain in very very short as:
--------------------------
- Moore's law will be not anymore true
so only memory will continue to increase ( static and volatil )
- Garbage Collector are not efficient in big memory for some 
technical reason
- Data to manage will continue to grow big data ant full memory 
stategy will the rule

So Develloper will move to a language where they are no garbage 
collector.
--------------------------

In bioinformatic we work with big data or full memory stategy 
often and that will not stop. So what think D garbage cllector's 
dev about this ?
Aug 28 2013
next sibling parent Jacob Carlborg <doob me.com> writes:
On 2013-08-28 23:28, bioinfornatics wrote:

 Hi everyone,

 yesterday i read an article into a french linux journal that in some
 years garbage collector will disapear.

 Why ? he explain in very very short as:
 --------------------------
 - Moore's law will be not anymore true
 so only memory will continue to increase ( static and volatil )
 - Garbage Collector are not efficient in big memory for some technical
 reason
 - Data to manage will continue to grow big data ant full memory stategy
 will the rule

 So Develloper will move to a language where they are no garbage collector.
 --------------------------

 In bioinformatic we work with big data or full memory stategy often and
 that will not stop. So what think D garbage cllector's dev about this ?


There has been some discussion about implementing ARC (Automatic 
Reference Counting) in the compiler.

-- 
/Jacob Carlborg
Aug 28 2013
prev sibling next sibling parent reply "qznc" <qznc web.de> writes:
On Wednesday, 28 August 2013 at 21:28:11 UTC, bioinfornatics 
wrote:

 Hi everyone,

 yesterday i read an article into a french linux journal that in 
 some years garbage collector will disapear.

 Why ? he explain in very very short as:
 --------------------------
 - Moore's law will be not anymore true
 so only memory will continue to increase ( static and volatil )
 - Garbage Collector are not efficient in big memory for some 
 technical reason
 - Data to manage will continue to grow big data ant full memory 
 stategy will the rule

 So Develloper will move to a language where they are no garbage 
 collector.
 --------------------------

 In bioinformatic we work with big data or full memory stategy 
 often and that will not stop. So what think D garbage 
 cllector's dev about this ?


Yes, I got this gist from hardware researchers repeatedly. Buzz 
phrases like "all computing is low power computing now" get 
thrown around. You can google "dark silicon" for more details.

However, your summary (or the article) is not quite correct. 
Garbage collection is not efficient in little-memory scenarios 
(embedded systems). Good, efficient garbage collection often 
requires a few times as much memory as manual memory management. 
This means if your application needs lots of RAM, you might be 
able tackle bigger tasks, if you get rid of the garbage collector.

Also, be careful about the problem domain. There will be enough 
domains left in some years, where garbage collection is the 
better tradeoff. Just think about all the domains where 
Python,Ruby,etc are popular now. The interesting field is mobile, 
though. Android's Dalvik and Web Apps use garbage collection. iOS 
switched to compiler-supported reference counting a few years 
ago. Mobile means saving battery is important, so efficiency is 
important. However, battery life is not the biggest factor for 
consumers.
Aug 29 2013
parent Jacob Carlborg <doob me.com> writes:
On 2013-08-29 16:20, qznc wrote:


 Also, be careful about the problem domain. There will be enough domains
 left in some years, where garbage collection is the better tradeoff.
 Just think about all the domains where Python,Ruby,etc are popular now.
 The interesting field is mobile, though. Android's Dalvik and Web Apps
 use garbage collection. iOS switched to compiler-supported reference
 counting a few years ago. Mobile means saving battery is important, so
 efficiency is important. However, battery life is not the biggest factor
 for consumers.


It's possible to use Ruby on iOS via developer tools similar to MacRuby. 
It uses automatic reference counting. Although, I've heard they having 
some problems with memory leaks, blocks in particular.

-- 
/Jacob Carlborg
Aug 29 2013
prev sibling next sibling parent "Brian Rogoff" <brogoff gmail.com> writes:
On Wednesday, 28 August 2013 at 21:28:11 UTC, bioinfornatics 
wrote:

 Hi everyone,

 yesterday i read an article into a french linux journal that in 
 some years garbage collector will disapear.

 Why ? he explain in very very short as:
 --------------------------
 - Moore's law will be not anymore true
 so only memory will continue to increase ( static and volatil )
 - Garbage Collector are not efficient in big memory for some 
 technical reason
 - Data to manage will continue to grow big data ant full memory 
 stategy will the rule


I've had similar thoughts myself for some time.


 So Develloper will move to a language where they are no garbage 
 collector.


There are many different kinds of software and developer, you 
need to qualify that assertion. GCed languages will not go away.



 --------------------------

 In bioinformatic we work with big data or full memory stategy 
 often and that will not stop. So what think D garbage 
 cllector's dev about this ?


I hope that D can go either way on this, and allows developers to 
'opt out' of GC if they need to, but I don't think it is that 
easy now. Is there a skeletal 'no GC D' project out there? What 
is given up (AAs, slices, classes, exceptions, ...) and how easy 
is it to use?

If not, Rust, Ada, and of course, C++ will be there. There are 
surprisingly few languages being designed to be used without GC 
these days.

-- Brian
Aug 29 2013
prev sibling parent "bioinfornatics" <bioinfornatics fedoraproject.org> writes:
so second question : I would like to know if using destroy 
function will help enough gc when working in big memory and 
saving time.
Aug 30 2013