11 Sep
2023
11 Sep
'23
8:53 p.m.
On Sep 11, 2023, at 2:09 PM, Martin Bishop via cctalk
<cctalk(a)classiccmp.org> wrote:
Correct.
Did Algol in general have memory leaks?,or is
just C and Windows.
This is very much from recollection of programming in Algol 60 & 68, 40+ years ago.
Algol60 did not have heap storage or pointers IIRC Therefore no potential for memory
leaks Algol68R had fairly agressive garbage collection,
which generally precluded memory leaks, in a language with dynamic arrays, ref ref ref
pointers, etc.
I don't know the 68R subset specifically. ALGOL 68 certainly has pointers, and heap
storage, where objects remain in existence so long as something refers to them.
That's similar to the model of BASIC strings, or Python objects. Garbage collection
or reference counting arethe usual way to deal with unreferenced objects, though things
get tricky when circular references are possible.
Also, the small memory model of initially 32 kiW
address space - which is all I can recollect ever using on an unpaged time sharing machine
ICL 1904S with (IIRC) 256 kiWd of physical memory - leads to a rather more static
programming style.
The grandchild of the Algols and this style of programming is perhaps subset Ada
A wag might observe that programmers may have memory leaks, that teams generate memory
leaks, and that organisations lack corporate memory.
:-)
The reason C and C++ have memory leaks is that they are low level languages where there is
no linkage between the lifetime of objects and the existence of references to them, nor in
fact a way for the implementation to know whether there are references. By constrast, in
strongly typed languages like ALGOL-68 or Pascal or Ada, and even in languages like
Python, that question does have a clear answer so memory management can be done by the
implementation rather than left to the tender mercies of the programmer.
paul