dlsspy

"Everybody" disagrees with you:

https://www.google.com/search?sourceid=chrome&ie=UTF-8&#...

That one's a bit tough since it really only affects people who are working far apart -- at which point you can just agree on what time something is.

For example, it's great to think of noon as being the middle of the day no matter where you are. There are clearly advantages the other way around as well.

DST is just disruptive, though. "Hey everybody! Let's have two ones o'clock this morning!" That created all kinds of confusion in my house last night when things were running at one I didn't expect to be running until two.

You spelled the name of the thing you're trying to abolish incorrectly. :/

Is there a petition for "make login work so I can sign petitions?"

This is bad because he was 0% productive for many days because he couldn't input text on the computer where the problem was because it didn't have his editor.

I bet you can. But can likely use emacs as well. I think if I had to pick one piece of software to live with on my desert island, it'd be emacs (assuming I had a reasonable boot loader and a filesystem to sit under it).

It took me years to consider using vim over plain vi. After fifteen years or so of vi I ended up switching to emacs.

I've worked with really awesome people who are completely lost without [insert proprietary single-platform editor here].

I am the 50%

SystemTap is not competition for dtrace. If you've ever been on a system that had dtrace, you'd find that pretty much everything works.

He's going through the same exercise I went through recently where your customer is running your software on Linux and you want to try to make use of the facilities offered by that OS so he can reproduce and diagnose the same issues.

The problem is that Linux just doesn't have good diagnostic tools. Most people don't recognize it because they've never had good tools. It's like being a C programmer who's never heard of a debugger. You can keep adding stuff to your code to try to isolate a bug and keep rerunning tests hoping it hits it where you've instrumented. That seems to be fine. But then you get a debugger and you can suddenly do significantly more on a running system gathering even more information on exactly the part that you care about.

This is what dtrace is. Linux needs something like it. If SystemTap will be that someday, then great. It has to work, though. My attempts to get it functional on Ubuntu and debian left me not feeling too great about the current state of thing (at least on debian, you can get kernel probes in without adding new and poorly documented package repository locations in). In neither case could I get SDTs working, though. For example, with dtrace, I can pretty trivially tell you the exact code paths in any running application on your system that are leading to IOPS. Then maybe I can exclude "known" paths and focus on only areas we don't think there should be any. This will include, for example, paging information. I can decide whether I want that when I'm drilling down. I can then look at another process that's running on the system and see when one is competing with the other for IOPS, CPU time, etc... I do this while the software is running and with minimal impact to the performance of the system.

When SystemTap does these things on my customer's Linux boxes, then I'd call it a dtrace competitor.

If you haven't decided, it's a great way to try it out.

I'm an opscode customer, but before that, I was playing around with their free plan.

The biggest annoyance is what it doesn't even look like it will do. I can't (meaningfully) sync my docs between my mac and my iPad. That's one of the main reasons I wanted to upgrade.

I brought up a thread on this on SO a while back. I wanted to find places where it didn't work:

http://stackoverflow.com/questions/328215/does-anyone-know-o...

It's significantly smaller when I do it on my machine with an optimizer.

Interestingly, g++ compiles it to the exact same code if you remove all the crap that isn't the main with the cout printing a literal string.

> Want to checkout all Apache Software Foundation projects? > svn checkout http://svn.apache.org/repos/asf/

Correction, that's all branches and all tags of all ASF projects. Not useful.

> Want to work on Apache httpd? > svn checkout http://svn.apache.org/repos/asf/httpd/

Again, all versions and all tags thereof.

> Maybe you only care about mod_proxy? > svn checkout http://svn.apache.org/repos/asf/httpd/httpd/trunk/modules/pr...

That's just the one version.

Similarly, my project wants the latest sources under under specific branches from repositories from five different administrative domains. That's trivial in git. I can move them around freely, lock a revision of a particular repository while letting others move forward, have custom (potentially unpublished or unwelcome upstream) patches to fit into my build system or supported OSes that are tracked along with them.

Subversion has externals for this same reason. It's no excuse for poor repository management.

Let me load the specially constructed set of parameters specific to this image so that when I do the next step you get a really clear image.

That was a little too hand-wavy. I'm a little dubious until I see what went into that phase.

Wow. That'd be fun to mix with labrea: http://dustin.github.com/2010/12/03/labrea.html

systemtap adds a ton of complexity over dtrace. You can do more with it, but the really simple things that have large benefits are no longer really simple (which causes them to lose some of their benefit).

I've burned a lot of time trying to get it working on ubuntu -- one of my guys actually did by bringing over a fedora kernel. He said it wasn't easy.

I'm hoping Linux gets something like this in the mainline. I don't care of it's dtrace, systemtap, or something in between. This functionality is critical for both development and operation of production systems. Unfortunately, people don't seem to understand that until they have access to such a tool and realize they can answer all kinds of questions they never knew they could.

I'll believe it when there's a buy now button.

Most advancements in technology feel like business as usual. Storage still freaks me out.

I wish people would compare these with IO operations instead of wall time. SSD is a huge win there, and that matters the most for a lot of people.

Security advice from someone making it easier to login as root on boxes...

The first things I do with sshd on any machine are 1) disable remote root login and 2) disable password login.

Looks neat. I could potentially make use of such a javascript library.

I like commit messages that capture the programmers thoughts well. A short, concise summary is important. The rest of the stuff is often things I want to know when I'm looking at a commit. Something is wrong with it, and I want to understand what the goal was as clearly as possible.

Doing more than one thing at once, no matter how trivial, often seems to bite, though.

> However, for the most part, the "efficiency" argument is off base. You can COMPILE regular expressions into tight code.

When I started programming ocaml, I was surprised to find no string primitives for things like split, find, etc... There was only regex. I didn't want to write a regex for doing a simple find or split, so I wrote a string library for my programs. My string library was consistently 4x faster than regex and did only exactly what I wanted.

Similarly, things that have configuration files that aren't in python, lua, etc... are significantly harder to use.

Go existed in various forms for a very long time. I was learning alef in the mid-nineties. When go came out, it was pretty easy to recognize.

That was a great explanation, though.

"write it again using new java 5 concurrent classes"

Java 5 was EOL'd in 2009. While it's good to understand some of the fundamentals (wait(long) is actually pretty tricky to use in practice), it'd be ridiculous not to allow anything introduced in the "new" java 5.

That's exactly how we use gerrit. The most recent change in our gerrit install is on its sixth revision:

   http://review.couchbase.org/8560

I pulled this one arbitrarily, it just happened to be the top. It's got a dependency and they change together. It works flawlessly and almost entirely transparently as long as you have a Change-Id (for which there's a simple hook to generate).

When I take my company out (~20-40 people depending on who's around), I need to first know whether the restaurant sells any food I can eat. If you make it hard for me to know that, I'm going to assume you don't and suggest we go either somewhere I know or some place that makes it easier for me to figure out.

Ubuntu 11.04 (GNU/Linux 2.6.38-10-generic x86_64)

It's mostly a (very low-volume) mail and web server. It's got ~200MB of web, about 11GB of "cache" files for a photo app I've got (about 31k files totally 11GB). I do a little bit of low volume S3 log processing on it every hour. My home directory is 2.5GB with about 107k files (that includes dropbox). The IMAP store is approaching 2 decades of mail with about 239k files totaling about 4.4GB of data. /root is a bit under 500MB with duplicity caches and stuff. Less than 300 files there.

I also keep a few snapshots I take every hour. There's a total of 347 snapshots across all the subvolumes (there are three).

    Filesystem            Size  Used Avail Use% Mounted on
    /dev/sda3             142G  102G   37G  74% /
    none                  2.0G  192K  2.0G   1% /dev
    none                  2.0G     0  2.0G   0% /dev/shm
    none                  2.0G   60K  2.0G   1% /var/run
    none                  2.0G     0  2.0G   0% /var/lock
    /dev/sda1             230M   69M  150M  32% /boot

Overall, it's a small atom box that lives a pretty boring life that I figured would have endless hard drive space given my patterns. That does not seem to be the case.

Anything further I can tell you?