Older blog entries for timj (starting at number 20)

Performance of a C++11 Signal System

C++11Signals

Performance of a C++11 Signal System

First, a quick intro for for the uninitiated, signals in this context are structures that maintain a lists of callback functions with arbitrary arguments and assorted reentrant machinery to modify the callback lists and calling the callbacks. These allow customization of object behavior in response to signal emissions by the object (i.e. notifying the callbacks by means of invocations).

Over the years, I have rewritten each of GtkSignal, GSignal and Rapicorn::Signal at least once, but most of that is long a time ago, some more than a decade. With the advent of lambdas, template argument lists and std::function in C++11, it became time for me to dive into rewriting a signal system once again.

So for the task at hand, which is mainly to update the Rapicorn signal system to something that fits in nicely with C++11, I’ve settled on the most common signal system requirements:

  • Signals need to support arbitrary argument lists.
  • Signals need to provide single-threaded reentrancy, i.e. it must be possible to connect and disconnect signal handlers and re-emit a signal while it is being emitted in the same thread. This one is absolutely crucial for any kind of callback list invocation that’s meant to be remotely reliable.
  • Signals should support non-void return values (of little importance in Rapicorn but widely used elsewhere).
  • Signals can have return values, so they should support collectors (i.e. GSignal accumulators or boost::signal combiners) that control which handlers are called and what is returned from the emission.
  • Signals should have only moderate memory impact on class instances, because at runtime many instances that support signal emissions will actually have 0 handlers connected.

For me, the result is pretty impressive. With C++11 a simple signal system that fullfils all of the above requirements can be implemented in less than 300 lines in a few hours, without the need to resort to any preprocessor magic, scripted code generation or libffi.

I say “simple”, because over the years I’ve come to realize that many of the bells and whistles as implemented in GSignal or boost::signal2 don’t matter much in my practical day to day programming, such as the abilities to block specific signal handlers, automated tracking of signal handler argument lifetimes, emissions details, restarts, cancellations, cross-thread emissions, etc.

Beyond the simplicity that C++11 allows, it’s of course the performance that is most interesting. The old Rapicorn signal system (C++03) comes with its own set of callback wrappers named “slot” which support between 0 and 16 arguments, this is essentially mimicking std::function. The new C++11 std::function implementation in contrast is opaque to me, and supports an unlimited number of arguments, so I was especially curious to see the performance of a signal system based on it.

I wrote a simple benchmark that just measures the times for a large number of signal emissions with negligible time spent in the actual handler.

I.e. the signal handler just does a simple uint64_t addition and returns. While the scope of this benchmark is clearly very limited, it serves quite well to give an impression of the overhead associated with the emission of a signal system, which is the most common performance relevant aspect in practical use.

Without further ado, here are the results of the time spent per emission (less is better) and memory overhead for an unconnected signal (less is better):

Signal System   Emit() in nanoseconds Static Overhead Dynamic Overhead
GLib GSignal 341.156931ns   0   0
Rapicorn::Signal, old  178.595930ns  64   0
boost::signal2   92.143549ns  24  400 (=265+7+8*16)
boost::signal   62.679386ns  40   392 (=296+6*16)
Simple::Signal, C++11    8.599794ns   8   0
Plain Callback    1.878826ns   -   -

 

Here, “Plain Callback” indicates the time spent on the actual workload, i.e. without any signal system overhead, all measured on an Intel Core i7 at 2.8GHz. Considering the workload, the performance of the C++11 Signals is probably close to ideal, I’m more than happy with its performance. I’m also severely impressed with the speed that std::function allows for, I was originally expecting it to be at least a magnitude larger.

The memory overhead gives accounts on a 64bit platform for a signal with 0 connections after its constructor has been called. The “static overhead” is what’s usually embedded in a C++ instance, the “dynamic overhead” is what the embedded signal allocates with operator new in its constructor (the size calculations correspond to effective heap usage, including malloc boundary marks).

The reason GLib’s GSignal has 0 static and 0 dynamic overhead is that it keeps track of signals and handlers in a hash table and sorted arrays, which only consume memory per (instance, signal, handler) triplet, i.e. instances without any signal handlers really have 0 overall memory impact.

Summary:

  • If you need inbuilt thread safety plus other bells and can spare lots of memory per signal, boost::signal2 is the best choice.
  • For tight scenarios without any spare byte per instance, GSignal will treat your memory best.
  •  If you just need raw emission speed and can spare the extra whistles, the C++11 single-file simplesignal.cc excels.

For the interested, the brief C++11 signal system implementation can be found here: simplesignal.cc
The API docs for the version that went into Rapicorn are available here: aidasignal.hh

PS: In retrospect I need to add, this day and age, the better trade-off for Glib could be one or two pointers consumed per instance and signal, if those allowed emission optimizations by a factor of 3 to 5. However, given its complexity and number of wrapping layers involved, this might be hard to accomplish.

Syndicated 2013-01-25 15:11:11 from Tim Janik

Sayebackup.sh – deduplicating backups with rsync

About

Due to popular request, I’m putting up a polished version of the backup script that we’ve been using over the years at Lanedo to backup our systems remotely. This script uses a special feature of rsync(1) v2.6.4 for the creation of backups which share storage space with previous backups by hard-linking files.
The various options needed for rsync and ssh to minimize transfer bandwidth over the Internet, time-stamping for the backups and handling of several rsync oddities warranted encapsulation of the logic into a dedicated script.

Resources

The GitHub release tag is here: backups-0.0.1
Script URL for direct downloads: sayebackup.sh

Example

This example shows creation of two consecutive backups and displays the sizes.

$ sayebackup.sh -i ~/.ssh/id_examplecom user@example.com:mydir # create backup as bak-.../mydir
$ sayebackup.sh -i ~/.ssh/id_examplecom user@example.com:mydir # create second bak-2012...-snap/
$ ls -l # show all the backups that have been created
drwxrwxr-x 3 user group 4096 Dez  1 03:16 bak-2012-12-01-03:16:50-snap
drwxrwxr-x 3 user group 4096 Dez  1 03:17 bak-2012-12-01-03:17:12-snap
lrwxrwxrwx 1 user group   28 Dez  1 03:17 bak-current -> bak-2012-12-01-03:17:12-snap
$ du -sh bak-* # the second backup is smaller due to hard links
4.1M    bak-2012-12-01-03:16:50-snap
128K    bak-2012-12-01-03:17:12-snap
4.0K    bak-current
Usage
Usage: sayebackup.sh [options] sources...
OPTIONS:
  --inc         make reverse incremental backup
  --dry         run and show rsync with --dry-run option
  --help        print usage summary
  -C <dir>      backup directory (default: '.')
  -E <exclfile> file with rsync exclude list
  -l <account>  ssh user name to use (see ssh(1) -l)
  -i <identity> ssh identity key file to use (see ssh(1) -i)
  -P <sshport>  ssh port to use on the remote system
  -L <linkdest> hardlink dest files from <linkdest>/
  -o <prefix>   output directory name (default: 'bak')
  -q, --quiet   suppress progress information
  -c            perform checksum based file content comparisons
  --one-file-system
  -x            disable crossing of filesystem boundaries
  --version     script and rsync versions
DESCRIPTION:
  This script creates full or reverse incremental backups using the
  rsync(1) command. Backup directory names contain the date and time
  of each backup run to allow sorting and selective pruning.
  At the end of each successful backup run, a symlink '*-current' is
  updated to always point at the latest backup. To reduce remote file
  transfers, the '-L' option can be used (possibly multiple times) to
  specify existing local file trees from which files will be
  hard-linked into the backup.
 Full Backups:
  Upon each invocation, a new backup directory is created that contains
  all files of the source system. Hard links are created to files of
  previous backups where possible, so extra storage space is only required
  for contents that changed between backups.
 Incremental Backups:
  In incremental mode, the most recent backup is always a full backup,
  while the previous full backup is degraded to a reverse incremental
  backup, which only contains differences between the current and the
  last backup.
 RSYNC_BINARY Environment variable used to override the rsync binary path.
See Also

Testbit Tools – Version 11.09 Release

flattr this!

Syndicated 2012-12-01 02:32:59 from Tim Janik

ListItemFilter Mediawiki Extension

For a while now, I’ve been maintaining my todo lists as backlogs in a Mediawiki repository. I’m regularly deriving sprints from these backlogs for my current task lists. This means identifying important or urgent items that can be addressed next, for really huge backlogs this can be quite tedious.

A SpecialPage extension that I’ve recently implemented now helps me through the process. Using it, I’m automatically getting a filtered list of all “IMPORTANT:”, “URGENT:” or otherwise classified list items. The special page can be used per-se or via template inclusion from another wiki page. The extension page at mediawiki.org has more details.

The Mediawiki extension page is here: http://www.mediawiki.org/wiki/Extension:ListItemFilter

The GitHub page for downloads is here: https://github.com/tim-janik/ListItemFilter

flattr this!

Syndicated 2012-11-23 17:58:17 from Tim Janik

Meeting up at LinuxTag 2012

 

Like every year, I am driving to Berlin this week to attend LinuxTag 2012 to attend the excellent program. If you want to meet up and chat about projects, technologies, Free Software or other things, send me an email or leave a comment with this post and we will arrange for it.

flattr this!

Syndicated 2012-05-15 13:12:56 from Tim Janik

Testbit Tools Version 11.09 Released

No Bugs
(Image: Mag3737)

 

And here’s another muffin from the code cake factory…

About Testbit Tools
The ‘Testbit Tools’ package contains tools proven to be useful during the development of several Testbit and Lanedo projects. The tools are Free Software and can be redistributed under the GNU GPLv3+.

This release features the addition of buglist.py, useful to aid in report and summary generation from your favorite bugzilla.

Downloading Testbit Tools
The Testbit Tools packages are available for download in the testbit-tools folder, the newest release is here: testbit-tools-11.09.0.tar.bz2

Changes in version 11.09.0:

  • Added buglist, a script to list and download bugs from bug trackers.
  • Added buildfay, a script with various sub commands to aid release making.
  • Fixed version information for all tools.
  • Added support to the Xamarin Bug Tracker to buglist.py.
  •  

    Feedback

    If you find this release useful, we highly appreciate your feature requests, bug reports, patches or review comments!

    See Also

    1. The Bugzilla Utility buglist.py – managing bug lists
    2. Wikihtml2man Introduction – using html2wiki

    flattr this!

    Syndicated 2011-10-01 00:18:55 from Tim Janik

    Wikihtml2man Introduction (aka html2man, aka wiki2man)

    Wiki↠HTML↠Man

     

    What’s this?
    Wikihtml2man is an easy to use converter that parses HTML sources, normally originating from a Mediawiki page, and generates Unix Manual Page sources based on it (also referred to as html2man or wiki2man converter). It allows developing project documentation online, e.g. by collaborating in a wiki. It is released as free software under the GNU GPLv3. Technical details are given in its manual page: Wikihtml2man.1.

    Why move documentation online?
    Google turns up a few alternative implementations, but none seem to be designed as a general purpose tool. With the ubiquituous presence of wikis on the web these days and the ease of content authoring they provide, we’ve decided to move manual page authoring online for the Beast project. Using Mediawiki, manual pages turn out to be very easily created in a wiki, all that’s then needed is a backend tool that can generate Manual Page sources from a wiki page. Wikihtml2man provides this functionality based on the HTML generated from wiki pages, it can convert a prerendered HTML file or download the wiki page from a specific URL. HTML has been choosen as input format to support arbitrary wiki features like page inclusion or macro expansion and to potentially allow page generation from other wikis than MediaWiki. Since wikihtml2man is based purely on HTML input, it is of course also possible to write the Manual Page in raw HTML, using tags such as h1, strong, dt, dd, li, etc, but that’s really much less convenient to use than a regular wiki engine.

    What are the benefits?
    For Beast, the benefits of moving some project documentation into an online wiki are:

    • We increase editability by lifting review requirements.
    • We are getting quicker edit/view turnarounds, e.g. through use of page preview functionality in wikis.
    • We allow assimilation of user contributions from non-programmers for our documentation.
    • Easier editability may lead to richer documentation and possibly better/frequently updated documentation.
    • Other projects also seem to make good progress by opening up some development parts to online web interfaces, like: Pootle translations, Transifex translations or PHP.net User Notes.

    What are the downsides?
    We have only recently moved our pages online and still need to gather some experience with the process. So far possible downsides we see are:

    • Sources and documentation can more easily get out of sync if they don’t reside in the same tree. We hope to be mitigating this by increasing documentation update frequencies.
    • Confusion about revision synchronization, with the source code using a different versioning system than the online wiki. We are currently pondering automated re-integration into the tree to counteract this problem.

    How to use it?
    Here’s wikihtml2man in action, converting its own manual page and rendering it through man(1):

      wikihtml2man.py http://testbit.eu/Wikihtml2man.1?action=render | man -l -

    Where to get it?
    Release tarballs shipping wikihtml2man are kept here: http://dist.testbit.eu/testbit-tools/.
    Our Tools page contains more details about the release tarballs.

    Have feedback or questions?
    If you can put wikihtml2man to good use, have problems running it or other ideas about it, feel free to drop me a line about it. Alternatively you can also add your feedback and any feature requests to the Feature Requests page (a forum will be created if there’s any actual demand).

    What’s related?
    We would also like to hear from other people involved in projects that are using/considering wikis to build production documentation online (e.g. in manners similar to Wikipedia). So feel free to leave a comment about your project if you do something similar.

    See Also

    1. New Beast Website – using html2wiki
    2. The Beast Documentation Quest – looking into documentation choices

    flattr this!

    Syndicated 2011-05-12 23:49:23 from Tim Janik

    Attending LinuxTag 2011

     

    Like every year, I am driving to Berlin this week to attend LinuxTag 2011 to attend the excellent program. If you want to meet up and chat about projects, technologies, Free Software or other things, send me an email or leave a comment with this post and we will arrange for it.

    Syndicated 2011-05-09 12:29:55 from Tim Janik

    BEAST v0.7.4 released

    BEAST/BSE version 0.7.4 is available for download at:

    BEAST is a music composition and modular synthesis application released as free software under the GNU LGPL that runs under Unix. Refer to the About page for more details.

    The 0.7.4 release integrates the bse-alsa package, several speedups, important bug fixes and translation updates.

    Please feel free to provide useful feedback or contribute on IRC, the mailing list and in the Wiki.

    TRANSLATORS: Please help us to improve the BEAST translation, just download the tarball, edit po/.po and email it to us or submit translations directly via the Beast page at Transifex.

    Overview of Changes in BEAST/BSE 0.7.4:

    • Renamed the project to Better Audio System / Better Sound Engine
    • Moved project website to: http://beast.testbit.eu/
    • Various build system fixes [stw,timj]
    • License fixups for some scripts [stw]
    • Fixed subnormal tests on AMD64 if SSE unit is in DAZ mode [stw]
    • Replaced slow resampler checks with a much faster resampling test [stw]
    • Performance improvements for various tests [stw]
    • GLib 2.28 unit test porting [stw]
    • Speed improvements for record field name [stw]
    • Fixed XRUNs in ALSA driver on 64bit systems [timj]
    • Added beast.doap [Jonh Wendell]
    • PO handling improvements.
    • Updated German translation.
    • Updated Norwegian bokmål translation [Kjartan Maraas]
    • Added e-Telugu translation [Veeven]

    flattr this!

    Syndicated 2011-04-09 01:50:16 from Tim Janik

    Human Multitasking

    Multitasking Mind
    (Image: Salvatore Vuono)

     

    The self deceiving assumption of effective human multitasking.

     

    People are often telling me they are good at multitasking, i.e. handling multiple things at once and performing well at doing so. Now, the human brain can only make a single conscious decision at a time. To understand this, we need to consider that making a conscious decision requires attention, and the very concept of attention means activating relevant information contexts for an observation or decision making and inhibiting other irrelevant information.

    The suppression involved in attention control makes it harder for us to continue with a previously executed task, this is why interruptions affect our work flows badly, such as an incoming call, SMS or a door bell. Even just making a decision on whether to take a call already requires attention diversion.

    Related, processing emails or surfing while talking to someone on the phone results in bad performance on both tasks, because the attention required for each, necessarily suppresses resources needed by the second task. Now some actions don’t suffer from this competition, we can walk and breathe or balance ourselves fine while paying full attention to a conversation. That’s because we have learned early on in our lives to automate these seemingly mundane tasks, so they don’t require our conscious attention at this point.

    Studies [1] [2] have shown time and again, that working on a single task in isolation yields vastly better results and in a shorter time frame when frequent context switches are avoided. This can be further optimized by training in concentration techniques, such as breath meditation, autogenic training or muscle relaxation.

    Here’s a number of tips that will help to put these findings to practical use:

    1. Let go of the idea of permanent reachability, nothing is so urgent that it cannot wait the extra hour to be handled efficiently.
    2. Make up your own mind about when to process emails, SMS, IM, news, voice messages.
    3. Start growing a habit of processing things in batches, e.g. walk through a list of needed phone calls in succession, compose related replies in batches, first queue and later process multiple pending reviews at once, queue research tasks and walk through them in a separate browsing session, etc.
    4. Enforce non-availability periods where you cannot be interrupted and may concentrate on tasks of your choice for an extended period.
    5. Schedule phone meetings in advance, ensure everyone has an agenda at hand for the meeting to avoid distractions (Don’t Call Me, I Won’t Call You).
    6. Deliberately schedule relaxation phases, e.g. take a 5 minute break off the screen per hour, ideally moving and walking around; rest breaks are needed after 90 minutes at latest.

    flattr this!

    Syndicated 2011-03-31 01:11:37 from Tim Janik

    Lanedo at CeBIT 2011

    This week, our people are running the Lanedo booth at CeBIT in Hannover.

    CeBIT Logo

    Everybody is invited to come and visit us in hall 2, booth D44/124, in the open source park. We will give introductions to our services, talk about current and future developments around GTK+ and Tracker and anything you want to approach us with.

    flattr this!

    Syndicated 2011-02-28 20:04:15 from Tim Janik

    11 older entries...

    New Advogato Features

    New HTML Parser: The long-awaited libxml2 based HTML parser code is live. It needs further work but already handles most markup better than the original parser.

    Keep up with the latest Advogato features by reading the Advogato status blog.

    If you're a C programmer with some spare time, take a look at the mod_virgule project page and help us with one of the tasks on the ToDo list!