Re: More on garbage collection

From: Nick Maclaren (nmm1_at_cus.cam.ac.uk)
Date: 06/13/05 

Date: 13 Jun 2005 16:36:15 GMT

In article <87d5qq495z.fsf-monnier+comp.arch@gnu.org>,
Stefan Monnier <monnier@iro.umontreal.ca> wrote:
>> |>
>> |> I wasn't just thinking of the area/zone/foo-based memory management
>> |> that existed back then, but more specifically of the more recent
>> |> variants where the deallocation is checked (either statically or
>> |> dynamically) to guarantee the "absence" of leaks and to prevent
>> |> dangling pointer dereference.
>
>> Yes, that is what I was doing then. It isn't hard.
>
>That's news to me. Could you point me to relevant material?

No. As I said, we regarded it as trivial and Just Did It. It is
only recently that people have been publishing claims that it is a
great new invention.

>> See my other postings. Difficulties in specifying contiguity,
>> alignment and so on, as well as in finding out how much space has
>> been used and is free. The problems redouble in spades when you
>> need to do either of these for physical memory rather than just
>> virtual.
>
>Sounds pretty vague to me. I mean I understand what is contiguity and
>alignment, but I'm not sure what you would consider as a solution.
>
>E.g. what do you actually use, and consider good memory
>management primitives for your HPC codes?

None that currently exist :-(

As a simple example, one thing that I should like to do is the following:

    Get a physically contiguous area of data of length N, aligned at
    2^K but not at the same 2^L alignment as pointers P and Q. And
    preserve those properties, swapping or cancelling the process
    rather than breaking them, if necessary.

It would be better to have a more portable description of such aspects,
but that would need more design than I have time for now.

Regards,
Nick Maclaren.

Relevant Pages

  • Re: More on garbage collection
    ... > |> dangling pointer dereference. ... That's news to me. ... > need to do either of these for physical memory rather than just ... alignment, but I'm not sure what you would consider as a solution. ...
    (sci.crypt)
  • Re: 1998 Suburban Alignment - Factor Pin Removal?
    ... The good news is that there is: the old string and chalk method can still ... The bad news is that alignment is a job that takes a whole lot of skill, ... competent operator to know what they mean and what to do. ... C'est suisse, et tres, tres precis." ...
    (rec.autos.tech)
  • Re: Alignment and format in listbox columns
    ... Regards, ... Claus ... > Alignment is set when you create the column header. ... > | how to control alignment for multiple columns. ...
    (microsoft.public.excel.programming)
  • Re: Uphill pull.
    ... Leroy F. Berven wrote: ... fervent ends, we could be for an interesting period of political re- ... alignment during the next decade or two. ... That's good news if it's true. ...
    (rec.arts.sf.fandom)
  • Re: "Generic" magic
    ... Same piss poor attitude and responce to reasoned arguments as over in ... rec.games.frp.dnd with regards to alignment. ...
    (rec.games.frp.gurps)