lynx-dev
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: lynx-dev Lynx 2.8.3.dev8: CPU problems connecting to a site


From: Klaus Weide
Subject: Re: lynx-dev Lynx 2.8.3.dev8: CPU problems connecting to a site
Date: Sun, 29 Aug 1999 13:51:53 -0500 (CDT)

On Sun, 29 Aug 1999, Vlad Harchev wrote:

>  Ok. (I'm satisfied by the fact that the bug is not in my code :).
>  Now I'm confused - what's happening? Can you check the virtual size of the
> lynx process - is it constantly increasing (if yes - then it's an endless loop
> in the lynx that allocates memory on each iteration), if no - then this is IMO
> bug in malloc - it iterates on the allocated chunks when searching for free
> ones. Is anybody with libc5 here (as in RH5.1 or Slackwares) (could you try to
> test this site with lynx configured that way)?
> 
>  Klaus, what do you think? It's your code (or you know it good) ..

It is nothing very mysterious (no need to assume anything is wrong with
malloc).   From Tom's CHANGES

      Add configure option --with-charsets to specify list
      of charsets which are not suppressed

Apparently, the way that lynx was compiled, "windows-1252" is not a
directly known charset.  In previous code version, it always[*] was
(unless one tweaked UCdomap and/or .tbl files by hand).

>  Why are invokations of UCGetLYhndl_byMIME (value=0x8146dbf "windows-1252")
> nested in each other? Why the nesting is not very deep?

When "windows-1252" is directly known (by being the MIMEname of a
registered character set), lookup of "windows-1252" succeeds directly;
and lookup of "cp1252" calls the same function recursively for
"windows-1252" which succeeds.

When "windows-1252" is not found directly, "cp1252" is looked up
instead by calling the same function recursively.  When "cp1252" is
not found directly, "windows-1252" is looked up instead.  The rest
follows...

[*] Well at some point in chartrans history, "windows-1252" was not known
(and then for a while more it was known under a different primary name).
But at that time, the recursive invocations were not as they are now, so
there was no possibility for this problem at that point.

  Klaus



reply via email to

[Prev in Thread] Current Thread [Next in Thread]