Hello,

While testing my following code:
https://github.com/ik5/redis_client.fpc/blob/master/tests/test_parser.lpr#L166

When I'm executing the code on the second item (that is suppose to be null in the protocol -> '$-1#13#10'),
the program crashes with the following message:
*** glibc detected *** redis_client/tests/test_parser: malloc(): memory corruption (fast): 0x0000000000750d50 ***

Using gdb to display the information seems that the following like: https://github.com/ik5/redis_client.fpc/blob/master/tests/test_parser.lpr#L85is the cause.
It looks like accessing either tmps or ALine[j] is causing it.

I can't figure out the reason for it.

I'm using FPC 2.6.0 x86_64 on Archlinux.

What am I missing here ?

Thanks,
Ido

I would expect that already line 83 is the problem. You have to change the order of the statements otherwise you'll access a not existing element in ALine before checking if the index is already to big when j becomes > alength. So change
while (ALine[j] #13) and (j <= alength) do
into
while (j <= alength) and (ALine[j] #13) do

But this will only work if the compiler switch {$B-} (verbose version {$BOOLEVAL OFF}) is set which is the default in Delphi so I assume the same in FPC too.
If you can't ensure that you should do it like this
while (j <= alength) do
  if (ALine[j]  #13) then

Regards

Michael

On Thu, Jan 26, 2012 at 09:10, Michael Müller <

Thanks,
I used valgrind and discovered this:
==20532== at 0x458754:
RD_TYPES_TREDISMULTIBULKRETURNTYPE_$__INSERT$LONGINT$TREDISRETURNTYPE (rd_types.pas:406)

It was off by one else where. Why does it take such time to effect the executable ?

Ido