On Wed, 02 Mar 2005 04:08:38 +, [EMAIL PROTECTED] [EMAIL PROTECTED] wrote:
demerphq [EMAIL PROTECTED] wrote:
:+ make_trie(startbranch,first,last,tail,flags)
snip
:+ This would be optimizable with startbranch=5, first=5, last=11, tail=16
'last=11' in the example args seems to contradict
On Wed, 2 Mar 2005 04:16:37 +, Dave Mitchell [EMAIL PROTECTED] wrote:
On Wed, Mar 02, 2005 at 04:08:38AM +, [EMAIL PROTECTED] wrote:
Hmm, locking a 'refcount--' doesn't make much sense to me: if two threads
both do this at the same time, they'll still then both try to free the
On Wed, 2 Mar 2005 04:16:37 +, Dave Mitchell [EMAIL PROTECTED] wrote:
On Wed, Mar 02, 2005 at 04:08:38AM +, [EMAIL PROTECTED] wrote:
Hmm, locking a 'refcount--' doesn't make much sense to me: if two threads
both do this at the same time, they'll still then both try to free the
demerphq wrote:
A trie is a way of storing keys in a tree structure where the
branching logic is determined by the value of the digits of the key.
Ie: if we have car, cart, carp, call, cull and cars we can
build a trie like this:
c + a + r + t
| | |
| | + p
| |
Is there any difference in the output from car, cart, carp, call,
cull
and cars; and from cart, carp, call, cull and cars (without car)?
Should the diagram below be amended to show that car is a terminal?
c + a + r +
| | |
| | + t
| | |
| | + p
|
On Tue, Mar 01, 2005 at 11:34:00PM -0600, [EMAIL PROTECTED] wrote:
Automated smoke report for 5.9.2 patch 23999
kirk.peters.homeunix.org: Intel(R) Celeron(R) CPU 2.00GHz (GenuineIntel
1994MHz) (i686/1 cpu)
onlinux - 2.6.10-1.766_FC3 [fedora]
using cc version 3.4.2
Dave Mitchell [EMAIL PROTECTED] wrote:
:On Wed, Mar 02, 2005 at 04:08:38AM +, [EMAIL PROTECTED] wrote:
: Hmm, locking a 'refcount--' doesn't make much sense to me: if two threads
: both do this at the same time, they'll still then both try to free the
: structure. But I see that op_free() does
demerphq [EMAIL PROTECTED] wrote:
:Hmm. Yes I think you are right. Ok. If i rework it so the trie is
:allocated and inserted into the regexes data structure at the get-go
:then pregfree would clean it up.
Make sure in that case that you can cope with reentrancy.
:Should i use distinct ids for
On Wed, 02 Mar 2005 14:53:21 +, [EMAIL PROTECTED] [EMAIL PROTECTED] wrote:
demerphq [EMAIL PROTECTED] wrote:
:Hmm. Yes I think you are right. Ok. If i rework it so the trie is
:allocated and inserted into the regexes data structure at the get-go
:then pregfree would clean it up.
Make
On Wed, 2 Mar 2005 12:26:24 -, Robin Barker [EMAIL PROTECTED] wrote:
Is there any difference in the output from car, cart, carp, call,
cull
and cars; and from cart, carp, call, cull and cars (without
car)?
Not structurally no. (more to come :-)
Should the diagram below be amended to
On Wed, Mar 02, 2005 at 05:20:36PM +0100, demerphq wrote:
What is the difference between the two sets? In my case should I use
ENTER/LEAVE or SAVETMPS/FREETMPS or both sets? Anybody who understands
these details who felt inclined to update the pod on this would find
me very grateful.
There
On Wed, 2 Mar 2005 20:44:35 +0100, demerphq [EMAIL PROTECTED] wrote:
'cars'=~/(cars|car)/ the order would be reversed even though the trie
Sorry I meant 'cars'=~/(cars|car)$/
For example for /car|cart|carp|call|cull|cars/ we would get
c -+- a -+- r.1 -+- t.2
| | |
|
-BEGIN PGP SIGNED MESSAGE-
Moin,
first let me say: I find that TRI stuff interesting. :) I know I have
written tons of code that does things like:
if ($type =~ /^(foo|bar|baz)$/)
{
...
}
One small question:
For example for
On Wed, 2 Mar 2005 19:46:43 +, Dave Mitchell [EMAIL PROTECTED] wrote:
On Wed, Mar 02, 2005 at 05:20:36PM +0100, demerphq wrote:
What is the difference between the two sets? In my case should I use
ENTER/LEAVE or SAVETMPS/FREETMPS or both sets? Anybody who understands
these details who
On Wed, Mar 02, 2005 at 09:52:46PM +0100, demerphq wrote:
Ok, so if im not using mortals, but rather SAVEFREEPV it looks like
the data is being stored on the save stack and not the tmpstack. So do
i still need the SAVETMPS/FREETMPS? Can I not just use ENTER/LEAVE by
itself?
yes. Or possibly
On Wed, 2 Mar 2005 21:22:55 +, Dave Mitchell [EMAIL PROTECTED] wrote:
On Wed, Mar 02, 2005 at 09:52:46PM +0100, demerphq wrote:
Ok, so if im not using mortals, but rather SAVEFREEPV it looks like
the data is being stored on the save stack and not the tmpstack. So do
i still need the
It may be a bug, but I'm not certainly not the best person
to try to fix it.
Problem is that the keys have binary names that may or
may not be valid UTF/Unicode characters. As near as I
can tell, there is no requirement on key or value _names_
other than that they cannot contain a \.
It would
perlfaq6.pod: /m allows caret and dollar to match next to a newline,
not just at the end of the string.
OK, add Remember also /g if you want to match more than once.
D Can you describe the code you were using that you found confusing?
$_=a\na\n; s/a/1/mg; print
D perlre (which the
D qx// ... should mention that the process inherits perl's stdin
D unless it's redirected.
Sounds good. Also include examples of the right way to do
$output=qx/command/EOF
input
EOF
instead of
$output=qx/commandEOF
input
EOF
/
and
[I have no idea]
instead of
$output=qx/echo $input|command/
i.e.,
On Wed, Mar 02, 2005 at 06:18:54 +0800, Dan Jacobson wrote:
instead of
$output=qx/echo $input|command/
i.e., how to input small things to qx without disrupting other parts
of the program or relying on /bin/sh constructions to get the input
connected.
I have a different idea... Since IPC::Run
demerphq [EMAIL PROTECTED] wrote:
:On Wed, 02 Mar 2005 14:53:21 +, [EMAIL PROTECTED] [EMAIL PROTECTED]
wrote:
: demerphq [EMAIL PROTECTED] wrote:
: :Hmm. Yes I think you are right. Ok. If i rework it so the trie is
: :allocated and inserted into the regexes data structure at the get-go
:
On Wed, Mar 02, 2005 at 10:37:51AM +0100, demerphq wrote:
BTW, should i send a more detailed high level explanation of what this
patch does? I assumed most people would be familiar with the ideas it
involves.
Someone eventually will need to write a one-paragraph version for
perldelta; might as
On Wed, Mar 02, 2005 at 04:08:38AM +, [EMAIL PROTECTED] wrote:
Hmm, that extra argument is used only if you've compiled perl with TRIE_DEBUG
(for which you have to patch the source), and are running with -Dr - that
seems like an unreasonable price for everyone else to pay.
I'd like to see
On Wed, Mar 02, 2005 at 05:20:36PM +0100, demerphq wrote:
Anyway, i suppose we can discuss this more later. I think for now hard
limits are a more reasonable approach.
How much memory would the array take for 10 |'d random 25-character
strings with a 4-character alphabet :)
Automated smoke report for 5.9.2 patch 23999
kirk.peters.homeunix.org: Intel(R) Celeron(R) CPU 2.00GHz (GenuineIntel
1994MHz) (i686/1 cpu)
onlinux - 2.6.10-1.766_FC3 [fedora]
using cc version 3.4.2 20041017 (Red Hat 3.4.2-6.fc3)
smoketime 5 hours 30 minutes (average 20
25 matches
Mail list logo
