From: Sebastian Gniazdowski <psprint@fastmail.com>
To: zsh-workers@zsh.org
Subject: Re: Possible huge setarrvalue optimization
Date: Fri, 18 Nov 2016 01:32:20 -0800 [thread overview]
Message-ID: <1479461540.1340250.791913609.27FAD722@webmail.messagingengine.com> (raw)
In-Reply-To: <1479449829.1305485.791811385.14DDFE28@webmail.messagingengine.com>
[-- Attachment #1: Type: text/plain, Size: 1417 bytes --]
Hello,
I attach patch that needs one problem to be resolved. Gains for
following test code:
test_fun() {
arr[20000]=""
repeat 10000; do
arr[10000]="x"
done
}
are huge: 37969 ms with no optimization, 244 ms with optimization. The
screensaver runs for 495 sec without optimization, 51 sec with opt
(still slower than hash which runs for 23 sec).
I changed:
if (v->end <= pre_assignment_length)
post_assignment_length += pre_assignment_length - v->end +
1;
to:
if (v->end <= pre_assignment_length)
post_assignment_length += pre_assignment_length - v->end;
And this allows to test post* == pre*. I think the "+ 1" isn't needed,
v->end can be treated as "number of elements affected already".
Subtracting the number of elements from pre_assignment_length gives
correct value: elements that will be copied from old.
The thing that has to be resolved: special setters. The patch contains
one special handler tested:
if ( pre_assignment_length != post_assignment_length ||
v->pm->gsu.a->setfn == set_region_active ) {
But this doesn't work as the symbol `set_region_active` isn't available.
How to solve this, and what are other special handlers? One test is
failing (A06) – append to typeset -U array, because the special setter
isn't called.
--
Sebastian Gniazdowski
psprint@fastmail.com
[-- Attachment #2: array_big_opt.diff --]
[-- Type: text/plain, Size: 1953 bytes --]
diff --git a/Src/params.c b/Src/params.c
index ef72cba..e23a0af 100644
--- a/Src/params.c
+++ b/Src/params.c
@@ -2654,24 +2654,34 @@ setarrvalue(Value v, char **val)
v->end = v->start;
post_assignment_length = v->start + arrlen(val);
- if (v->end <= pre_assignment_length)
- post_assignment_length += pre_assignment_length - v->end + 1;
-
- p = new = (char **) zshcalloc(sizeof(char *)
- * (post_assignment_length + 1));
-
- for (i = 0; i < v->start; i++)
- *p++ = i < pre_assignment_length ? ztrdup(*q++) : ztrdup("");
- for (r = val; *r;) {
- /* Give away ownership of the string */
- *p++ = *r++;
- }
- if (v->end < pre_assignment_length)
- for (q = old + v->end; *q;)
- *p++ = ztrdup(*q++);
- *p = NULL;
- v->pm->gsu.a->setfn(v->pm, new);
+ if (v->end <= pre_assignment_length)
+ post_assignment_length += pre_assignment_length - v->end;
+
+ if ( pre_assignment_length != post_assignment_length || v->pm->gsu.a->setfn == set_region_active ) {
+ p = new = (char **) zshcalloc(sizeof(char *)
+ * (post_assignment_length + 1));
+
+ for (i = 0; i < v->start; i++)
+ *p++ = i < pre_assignment_length ? ztrdup(*q++) : ztrdup("");
+ for (r = val; *r;) {
+ /* Give away ownership of the string */
+ *p++ = *r++;
+ }
+ if (v->end < pre_assignment_length)
+ for (q = old + v->end; *q;)
+ *p++ = ztrdup(*q++);
+ *p = NULL;
+
+ v->pm->gsu.a->setfn(v->pm, new);
+ } else {
+ /* v->start is 0-based */
+ p = old + v->start;
+ for (r = val; *r;) {
+ /* Give away ownership of the string */
+ *p++ = *r++;
+ }
+ }
/* Ownership of all strings has been
* given away, can plainly free */
[-- Attachment #3: testopt7.zsh --]
[-- Type: application/octet-stream, Size: 242 bytes --]
#!/usr/local/bin/zsh-arr-assign
#!/usr/local/bin/zsh-arr-assign-before
zmodload zsh/zprof
typeset -a arr
test_fun() {
arr[20000]=""
repeat 10000; do
arr[10000]="x"
done
}
test_fun
print -rl $arr[@] $arr[10000]
zprof
next prev parent reply other threads:[~2016-11-18 9:32 UTC|newest]
Thread overview: 13+ messages / expand[flat|nested] mbox.gz Atom feed top
2016-11-18 6:17 Sebastian Gniazdowski
2016-11-18 7:09 ` Sebastian Gniazdowski
2016-11-18 9:32 ` Sebastian Gniazdowski [this message]
2016-11-18 12:20 ` Sebastian Gniazdowski
2016-11-20 11:46 ` Daniel Shahaf
2016-11-20 17:41 ` Bart Schaefer
2016-11-20 20:54 ` Sebastian Gniazdowski
2016-11-20 21:19 ` Peter Stephenson
2016-12-24 17:19 ` Daniel Shahaf
2017-01-04 18:31 ` Sebastian Gniazdowski
2017-01-05 4:13 ` Daniel Shahaf
2017-01-05 13:32 ` Sebastian Gniazdowski
2017-01-05 15:47 ` Daniel Shahaf
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=1479461540.1340250.791913609.27FAD722@webmail.messagingengine.com \
--to=psprint@fastmail.com \
--cc=zsh-workers@zsh.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
Code repositories for project(s) associated with this public inbox
https://git.vuxu.org/mirror/zsh/
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).