Tim Bunce [Wed, 17 Oct 2012 09:46:25 +0000]
fix bin/sizeme_graph.pl for older Mojolicious versions RT80195
Tim Bunce [Wed, 17 Oct 2012 09:25:52 +0000]
fix madprop_size, hopefully (untested)
Tim Bunce [Wed, 17 Oct 2012 09:18:41 +0000]
fix pod NAME
Tim Bunce [Thu, 11 Oct 2012 15:34:50 +0000]
Bump to 0.05
Tim Bunce [Thu, 11 Oct 2012 15:34:09 +0000]
fix gettimeofday_nv proto
Tim Bunce [Thu, 11 Oct 2012 15:33:33 +0000]
ifdef PL_my_cxt_size
Tim Bunce [Thu, 11 Oct 2012 15:17:13 +0000]
use PERL_BCDVERSION for SVt_LAST
Tim Bunce [Thu, 11 Oct 2012 15:06:24 +0000]
add .tar and .bak to .gitignore
Tim Bunce [Thu, 11 Oct 2012 15:05:37 +0000]
fix compiler warning
Tim Bunce [Thu, 11 Oct 2012 15:00:49 +0000]
Merge branch 'tim-
20120930-sizeme' of github.com:timbunce/devel-sizeme into tim-20120930-sizeme
Tim Bunce [Thu, 11 Oct 2012 15:00:16 +0000]
fix madprop_size declaration
Tim Bunce [Thu, 11 Oct 2012 14:54:18 +0000]
Tweak PERL_BCDVERSION for ps->compcv
Tim Bunce [Thu, 11 Oct 2012 14:52:25 +0000]
Fix thread issues per cpantesters.org/cpan/report/
e0aa7316-118c-11e2-8948-fc6d58678c7a
Tim Bunce [Thu, 11 Oct 2012 14:45:01 +0000]
Use Devel::Dwarn instead of Data::Dump
Tim Bunce [Thu, 11 Oct 2012 14:02:16 +0000]
Merge pull request #1 from letolabs/fix_clang_warnings
Fix compiler warnings from clang 2.1 thanks to leto
Tim Bunce [Thu, 11 Oct 2012 14:00:58 +0000]
Bump VERSION to 0.04. Tweak a warning message
Jonathan "Duke" Leto [Wed, 10 Oct 2012 20:23:39 +0000]
Fix compiler warnings from clang 2.1
This fixes the following warnings:
SizeMe.xs:273:44: warning: conversion specifies type 'int' but the argument has type 'I32' (aka 'long')
SizeMe.xs:1133:50: warning: conversion specifies type 'int' but the argument has type 'U32'
Tim Bunce [Fri, 5 Oct 2012 16:48:30 +0000]
Removed Mojolicious::Lite as a prerequisite as it was stoping people installing on 5.8
Tim Bunce [Fri, 5 Oct 2012 16:27:15 +0000]
Added CHANGES for 0.03
Tim Bunce [Fri, 5 Oct 2012 12:39:01 +0000]
Polish docs and bump version for 0.03.
Tim Bunce [Thu, 4 Oct 2012 17:42:58 +0000]
Skip nest invalidated by 5.17 change
Tim Bunce [Thu, 4 Oct 2012 09:00:57 +0000]
Bump VERSION to 0.020_082
Tim Bunce [Thu, 4 Oct 2012 08:47:35 +0000]
No longer chase CvSTART (dubious before 5.17 and broken with it)
Tim Bunce [Thu, 4 Oct 2012 08:08:33 +0000]
Fix COPHH for PERL_BCDVERSION < 0x5013007
Tim Bunce [Wed, 3 Oct 2012 13:51:40 +0000]
bump version
Tim Bunce [Wed, 3 Oct 2012 13:50:49 +0000]
Fix for coredumped in t/recurse.t with a non-threaded 5.8.9
Tim Bunce [Wed, 3 Oct 2012 13:41:07 +0000]
Fix most backcompat issues (tested with 5.8.9)
Tim Bunce [Wed, 3 Oct 2012 12:55:45 +0000]
PL_apiversion PL_ofsgv PL_custom_ops. basecop->cop_stashpv commented out for now.
Tim Bunce [Wed, 3 Oct 2012 12:41:01 +0000]
parser compcv vs comppad
Tim Bunce [Wed, 3 Oct 2012 12:26:50 +0000]
avoid COPHH and define CopHINTHASH_get if not defined
Tim Bunce [Wed, 3 Oct 2012 12:10:37 +0000]
Fix pod coverage.
Tim Bunce [Wed, 3 Oct 2012 11:56:21 +0000]
tweak .gitignore
Tim Bunce [Wed, 3 Oct 2012 11:31:59 +0000]
applied ppport.h recommended patch
Tim Bunce [Wed, 3 Oct 2012 11:31:03 +0000]
update ppport.h to 3.20
Tim Bunce [Wed, 3 Oct 2012 11:27:52 +0000]
Update MANIFEST. Tweak VERSION for dev release.
Tim Bunce [Tue, 2 Oct 2012 21:40:01 +0000]
Fix sizeme_store.pl pod typo. Make -d:SizeMe be special only if it's first module.
Tim Bunce [Tue, 2 Oct 2012 21:38:58 +0000]
Move sizeme_graph.pl to bin and the static files into lib so they're installed.
Tim Bunce [Tue, 2 Oct 2012 20:11:25 +0000]
Polish sizeme_graph.pl
Tim Bunce [Tue, 2 Oct 2012 19:54:14 +0000]
Only set SIZEME for -d mode. Fix assertion when SIZEME env var not set.
Tim Bunce [Tue, 2 Oct 2012 19:28:35 +0000]
More docs and other polish for release.
Tim Bunce [Tue, 2 Oct 2012 19:28:10 +0000]
Tweak tooltip.
Tim Bunce [Tue, 2 Oct 2012 19:27:16 +0000]
Merge any node with a single child. Added --showid and more --debug output.
Tim Bunce [Tue, 2 Oct 2012 16:34:00 +0000]
Remove 'pre attr'. Other minor cleanup.
Tim Bunce [Tue, 2 Oct 2012 16:27:42 +0000]
renaming for static/sizeme_graph.pl
Tim Bunce [Tue, 2 Oct 2012 15:32:54 +0000]
Merge remote-tracking branch 'origin/rafl' into tim-
20120930-sizeme
Conflicts:
.gitignore
Makefile.PL
SizeMe.xs
Also added #ifdef PERL_IMPLICIT_CONTEXT around PL_my_cxt_size.
Tim Bunce [Tue, 2 Oct 2012 14:54:31 +0000]
Rename to Devel::SizeMe
Tim Bunce [Tue, 2 Oct 2012 14:11:49 +0000]
Assorted minor changes.
Florian Ragwitz [Mon, 1 Oct 2012 09:22:13 +0000]
Count all stack frames, not just the used ones
Florian Ragwitz [Mon, 1 Oct 2012 09:21:01 +0000]
Fix a bunch of warnings
Florian Ragwitz [Mon, 1 Oct 2012 06:13:12 +0000]
Add a few more global pointers
Florian Ragwitz [Mon, 1 Oct 2012 02:51:12 +0000]
Add more interpreter globals
Florian Ragwitz [Mon, 1 Oct 2012 02:19:13 +0000]
Count PerlInterpreter when applicable
Florian Ragwitz [Mon, 1 Oct 2012 02:04:16 +0000]
For PL_compiling, don't count the actual OP
It's either stored in the data section or in the PerlInterpreter struct, so it
should not be counted or counted as part of struct PerlInterpreter. However, we
do wanna count things referenced from PL_compiling, such as cop_file and the
hints hash, if present.
Florian Ragwitz [Mon, 1 Oct 2012 02:03:22 +0000]
Allow the op structure to not be counted
Depending on how the OP is allocated, you may or may not want to count the size
of the actual OP. However, you'll want to count other data referenced from the
OP.
Florian Ragwitz [Mon, 1 Oct 2012 01:49:45 +0000]
Add a few more globals
Some of them were already reached on other paths, others weren't.
Florian Ragwitz [Mon, 1 Oct 2012 01:31:38 +0000]
Always count PL_compiling as a COP
It's always allocated as a full COP, never as anything else that could possibly
be smaller. Pass that information along.
Florian Ragwitz [Mon, 1 Oct 2012 01:29:16 +0000]
Factor out op class guessing
Sometimes callers already know what kind of class the op they're dealing with
belongs to. If they do, they can now just pass down that information into
op_size_class, avoiding possibly wrong guesses from cc_opclass based on op_type.
Florian Ragwitz [Mon, 1 Oct 2012 01:27:16 +0000]
Attempt to count PL_compiling
Often, PL_compiling will have an op_type of OP_NULL, making it being counted as
a regular base op. However, it's memory is being allocated within the
PerlInterpreter structure or in a proper global as a full COP, not just a
BASEOP. Therefore it's possible for it to be counted wrong right now.
Florian Ragwitz [Mon, 1 Oct 2012 01:09:20 +0000]
Count lexical hint hashes in COPs
Florian Ragwitz [Mon, 1 Oct 2012 01:07:29 +0000]
Add refcounted_he structs counting
Florian Ragwitz [Mon, 1 Oct 2012 01:05:32 +0000]
Attempt to extract the refcounted_he definition
Not all perls have this structure, so some compatibility checks will have to be
added later.
We can't easily reuse the definition from hv.h unless we define PERL_CORE, which
will break loads of other code we currently have.
Tim Bunce [Sun, 30 Sep 2012 09:49:27 +0000]
make hash keys be a he+hek node that has the value as a child
Tim Bunce [Sun, 30 Sep 2012 09:48:07 +0000]
Mark suspect PERL_SUBVERSION macros
Tim Bunce [Sun, 30 Sep 2012 09:16:58 +0000]
tweak --showid output
Tim Bunce [Sun, 30 Sep 2012 09:09:10 +0000]
Rework AV index labels to avoid using NPattr_PRE_ATTR
Tim Bunce [Sun, 30 Sep 2012 08:18:34 +0000]
add external gz mechanism as experiment (~3 pct)
Tim Bunce [Sun, 30 Sep 2012 08:16:48 +0000]
Show duration of SizeMe streaming
Tim Bunce [Sun, 30 Sep 2012 08:16:37 +0000]
cope with array index name attr being undef
Tim Bunce [Sun, 30 Sep 2012 08:14:41 +0000]
Make sv_size return true if sv not seen before.
Florian Ragwitz [Sun, 30 Sep 2012 08:03:12 +0000]
Add a missing dependency
Tim Bunce [Sun, 30 Sep 2012 07:55:37 +0000]
Add S & E tokens, plus timing.
Florian Ragwitz [Sun, 30 Sep 2012 07:50:52 +0000]
Move SvSTASH counting from SVt_PVCV to >= SVt_PVMG
I assume that certain stashes might not be reached using the previous
behaviour. Both objects and CVs (and possibly even others) can have a stash.
For anonymous stashes without CVs in them, I believe SvSTASH would not be
counted if an object blessed in said anonymous stash would be processed.
Florian Ragwitz [Sun, 30 Sep 2012 07:50:12 +0000]
For non-magical XPVMGs, count OURSTASH for if present
Florian Ragwitz [Sun, 30 Sep 2012 07:48:30 +0000]
Only attempt to count magic on magical SVs
XPVMGs might not have a MAGIC* chain attached. SvMAGICAL is the predicate for
that.
Florian Ragwitz [Sun, 30 Sep 2012 07:31:33 +0000]
Fix compilation on perls with IMPLICIT_CONTEXT
Tim Bunce [Sun, 30 Sep 2012 06:56:27 +0000]
First steps migrating to SizeMe
Tim Bunce [Fri, 28 Sep 2012 19:19:03 +0000]
More prereqs
Tim Bunce [Fri, 28 Sep 2012 19:18:06 +0000]
Merge branch 'tim-
20120929-d-m' of git.shadowcat.co.uk:Devel-Size into tim-20120929-d-m
Tim Bunce [Fri, 28 Sep 2012 19:16:53 +0000]
Don't "open -a Graphviz $opt_dot" unless on OSX
Tim Bunce [Fri, 28 Sep 2012 19:16:22 +0000]
tweak .gitignore
Tim Bunce [Fri, 28 Sep 2012 19:14:56 +0000]
add some PREREQ_PM
timbo [Fri, 28 Sep 2012 18:57:04 +0000]
Fix mstats scope
Tim Bunce [Fri, 28 Sep 2012 18:54:39 +0000]
mstats() isn't very portable
Tim Bunce [Fri, 28 Sep 2012 18:31:11 +0000]
Mega rename to Devel::Memory commit
Tim Bunce [Fri, 28 Sep 2012 16:09:57 +0000]
add TODO
Tim Bunce [Fri, 28 Sep 2012 16:08:08 +0000]
Split out a new heap_size() function. Add TODO. Special case PL_strtab HeVAL's.
Tim Bunce [Fri, 28 Sep 2012 07:35:26 +0000]
tweak Children tooltip
Tim Bunce [Fri, 28 Sep 2012 07:30:53 +0000]
Add "->" to link names for --text.
Tim Bunce [Fri, 28 Sep 2012 01:51:20 +0000]
Add logarea param via referer. add bytesToSize. Polish formatting of tooltip
Tim Bunce [Thu, 27 Sep 2012 16:09:47 +0000]
update notes.txt
Tim Bunce [Thu, 27 Sep 2012 15:27:24 +0000]
Hack to enable log scale for treemap nodes (disabled)
Tim Bunce [Thu, 27 Sep 2012 15:24:21 +0000]
Fix for scalar attr and assorted clenups
Tim Bunce [Thu, 27 Sep 2012 15:13:11 +0000]
check for AVelem, remove Dwarn
Tim Bunce [Thu, 27 Sep 2012 15:12:43 +0000]
tweaks
Tim Bunce [Wed, 26 Sep 2012 17:14:32 +0000]
Add PRE_ATTR and use it for array indices.
Tim Bunce [Wed, 26 Sep 2012 17:09:07 +0000]
Apply SOME_RECURSION to CvOUTSIDE for now
Tim Bunce [Wed, 26 Sep 2012 17:08:58 +0000]
Add 'ADD_PRE_ATTR' mechanism so we can output attrs for the next node.
Tim Bunce [Wed, 26 Sep 2012 16:06:06 +0000]
Using check_new doesn't work for padlist_size. Mark weakrefs in link name.
Tim Bunce [Wed, 26 Sep 2012 15:09:52 +0000]
Use check_new in padlist_size.
Tim Bunce [Wed, 26 Sep 2012 15:09:12 +0000]
Pad names almost working. Some asorted code cleanup.