John Sequeira

Chris Winters
Lars Pind
Dan Sugalski
Ask Bjoern Hansen
Emacs
Phil Windley
Wes Felter
Jon Udell
Sanjay

blog

Amped::Technology

Thursday, May 26, 2005
EnterpriseDB: ora2pg and mssql2pg?

From the EnterpriseDB web site:

We have worked hard to make EnterpriseDB "feel" like the enterprise databases you're used to. Its built-in “Redwood” compatibility mode will be comfortable for long-time developers and DBAs, and it will work with many existing applications as well. “Redmond” mode is on the way. And EDB’s tools look and behave the way you expect them to.

In their press coverage, E-DB has alluded that they'll be shipping run-time compatibility for Oracle and SQL Server, and this web site snippet is further confirmation. (I hope that's what they mean, and not that they're simply skinning the admin tools.) But I wonder how compelling a feature this will be, given that Ingres already has design-time migration tools, and doesn't seem to be generating much buzz.

Anyway, I was curious enough about the product to have set up a call tomorrow with EnterpriseDB's CEO Andy Astor. If you have any questions you'd like me to ask about the forthcoming release, please leave a comment or drop me an email. More to come.
9:49:56 PM      comment []  trackback []


Wednesday, May 25, 2005
Apache Foundation Meets Solaris Zones.

Stefano Mazzocchi lives about 2 blocks from me in Cambridge, MA. I don't run into him very often, but I always learn alot when I do. A month or two ago, he was describing some challenges facing the Apache Foundation. He mentioned that the the apache.org sysadmins are extraordinarily paranoid about defending their high-value target, due to a successful cvs compromise a few years ago. The paranoia, while justified, has really held up the functionality they could offer incubator projects (arbitrary cgi was a no-no) and prevented any delegated administration which led to a serious bottleneck on admin time. I'm sure I asked about using virtualization to sandbox the various projects, and he said they were looking into it.

Gregor (who's one of the Apache lenya authors) recently was able to use Apache's Solaris Zones environment to create an online demo of his cms.... so it looks like they're officially on the virtualization boat. Nice

For extra credit, calculate how much money they saved not using Windows to accomplish this.
1:13:09 AM      comment []  trackback []


Tuesday, May 24, 2005
Stefano's Piggybank: Distributed Tagsonomy

Stefano released a new version of Piggybank today, his semantic web killer app. Two interesting items left out of Stefano's post:

  1. Piggybank allows you to run private 'banks' for non-public annotation of web resources. The simplest way to think of this is like a distributed de.licio.us -- one in which you own and control your annotations, but can share them with whoever you like. I think that's a model worth emulating.
  2. Piggybank ships as a firefox plugin but runs as an out-of-process web server. At ~15Meg or so (sans Java runtime), I think it takes the crown for most-sizable firefox extension. :-)

(Thanks to Ken Novak for the synopsis)
2:02:36 PM      comment []  trackback []


Friday, May 20, 2005
The Answer: Cheap Gas and Chicago Crime

The Question: What are two nifty examples of maps.google.com visualization?

Thanks for playing blogger Jeopardy
12:22:18 PM      comment []  trackback []


Thursday, May 19, 2005
The OS is Dead, Long Live the Hypervisor

The implications of such virtualization are enormous. For one, you would be able to run multiple operating systems on your desktop. Perhaps you have wanted to try the free version of Pro Tools that only runs on Windows 98 or would love to add a light Linux-based CAD program like CADvas to your system. No problem. The operating systems will not interfere with each other or the applications.

... In this paradigm, the OS could not be less important except as a tool to run the applications you need.

The article discusses using hypervisors as a replacement for the operating system, and application-specific VMs as opposed to the hassle of maintaining software on a single system. I've certainly been thinking about this for a few years, probably due to the fact that I have to work with so many different server platforms, and virtualization has enabled me to do this without requiring the sysadmin skills that would otherwise be necessary. (Ah, laziness).

It's interesting to consider how software would be packaged and delivered in a world like this: what software/etc becomes possible if installation, versioning, conflicts, etc goes away, and you essentially just sit down in front of a machine (with your old data, of course) dedicated to just that task or set of tasks.

Lot's of digerati talk about the novelty of your data being in the cloud and untethered to an actual physical location. I think it gets much more interesting when you couple that with the idea that the machine or v-machines you do your work on would also live in the cloud. They could follow you around either via remote access or by synchronizing big files -- both problems for which pretty mature solutions exist. Your data then doesn't have to leave your house if for security or privacy reasons you don't want it to. Executing as an ASP could be reduced to simply maintaining a VM for download or rsync by your customers, instead of dealing with server farms, deployment etc.

Issues exist, like security/viruses, desktop integration, customization, etc, but since the vm-vs-single-host paradigms aren't mutually exclusive (in fact they're complementary to a degree almost never seen by disruptive tech) I don't see any of the issues slowing the momentum towards a hypervisor world.
12:35:05 PM      comment []  trackback []


Friday, May 13, 2005
Edward Tufte Meets Baby Names

Not necessarily general interest, but a pretty cool visualization nonetheless.
1:10:39 PM      comment []  trackback []

Wednesday, May 11, 2005
ROR in Javascript?

"TrimJunction is a clone of Ruby on Rails, into JavaScript."
10:02:35 PM      comment []  trackback []
Microsoft's oracle2mssql

"With a single click, SSMA converts all database objects, including stored procedures, functions, triggers, views, packaged procedures, packaged functions, and sequences. SSMA also provides context-sensitive warnings and output messages. The conversion engine encapsulates Microsoft's patent-pending technology that represents one of the most complete PL/SQL to T-SQL conversion capabilities available."

9:56:43 PM      comment []  trackback []

Friday, May 06, 2005
In-situ Web Testing

Write tests for your web app using javascript?

Over the years, my webtest tool of choice has migrated from the declarative WWW::Chat and HTTP::WebTest to Test::WWW::Mechanize.

Although I prefer the declarative approach for testing, it seems like there is always some procedural code involved and patching the two approaches together is always awkward. Also, the Mech* modules have the biggest following right now (and the best docs etc.) and correspondingly have become the easiest to use.

The tools I use do not let you run the browser engine on pages, however (evaluating javascript etc.) and have limited usefulness testing ajax-y types of pages. I think I will revisit thoughtworks when I have that need.
9:29:12 AM      comment []  trackback []

Search Serendipity

I've been using A9 for the last few months, and I quite like the experience. When I switched, I was trying to figure out whether there was any value to be had mucking up the Google interface with sticky search (bookmarks, diary, search history etc). Mostly the answer was no, as I haven't gotten much value out of those options. But with the new open search api, you can create your own metasearches. I like this a lot.

It's has an effect pretty similar to defining your own text ads to run down the side of your search results. In my case, since I'm working on several perl projects at the moment, I've added the CPAN search widget, so that search.cpan.org results appear alongside my google results. This is useful because a lot of what I search for is perldoc-type stuff (where search.cpan results would be relevant), but even when it's not directly what I'm looking for it exposes me to random perl modules somewhat similar to what I'm looking for (which is a experience akin to glancing at nearby shelves in a library.)

Although I'm not really using it that way, picking your own ads seems like a good way to bypass the otherwise balkanized nature of net media.
9:08:53 AM      comment []  trackback []


Thursday, May 05, 2005
You can Fiddle

with firefox, too.
9:29:56 AM      comment []  trackback []
Amped::On::IO::All

Here's a code snippet that uses CPAN module IO::All to return a pdf of a dynamic HTML page to the browser. IO::All demonstrates the potential for API simplication w/operator overloading when applied to something mundane like file access:

    my $fname = ('?'); # create temp file handle
    my $htmldoc = io("| $htmldoc  -t pdf14 -f $fname --datadir $htmldoc_dir --footer t/D  -")   
      or $logger->fatal( "Couldn't open htmldoc: $! $?" );
    $html > $htmldoc;  # pass string to pipe
    $htmldoc->close();
        
    my $pdf = io($fname);    
    my $binary = $pdf->binary->scalar; # slurp up output pdf file
    $pdf->unlink();  #delete temp file
    $pdf->close();
    
    return $binary;

IO::All author Ingy has done some really good work smoothing over the rough corners of perl. If that's something that interests you, I suggest looking over two of his other modules: Spiffy and YAML. The ajax-enthusiasts are discovering the YAML-subset JSON ... so maybe it will get some of the popularity it deserves.
9:24:51 AM      comment []  trackback []

Why Tables for Layout is Stupid

This presentation may be a couple of years old, but some of the smartest programmers I know (myself included!) never quite got around to taking a look at CSS for layout.

I'm not much of a 'front-end' guy, but sometimes it has to be done. Justification + some helpful hints on doing it right (i.e. w/css).

(via Tucows Developer Hangout)
9:21:40 AM      comment []  trackback []


Monday, May 02, 2005
Joel on Software on VMware

We bought high-end, dual-Xeon Dell workstations with dual monitors and 2 GB of memory each (that's our new standard developer workstation) which will make it easy and fast to simulate the complete production environment on every intern's computer. It's probably overkill and you can run VMware happily on less pumped-up computers, but if you're simulating a network of three machines VMware definitely benefits from lots of RAM and CPUs.

I've got a similar vmware-tuned rig: dual athlon w/2g RAM, and I'm actually finding it to be overkill. VMWare Workstation 5 stole an important memory optimization from ESX and GSX server. It keeps a single copy of memory pages that are identical in different VM instances, so running three pristince copies of W2K eats up the RAM of just one copy (less if your host is running the same OS). I'm running an IIS instance with 384M and a SQL Server with the same, and task manager shows my total system memory (including host running emacs/firefox/word/thunderbird/vnc/logmein/prevx/spybot/radiouserland/etc) at 532Meg. With 5.0, I expect it'll be a long time before I tap into that second Gig of RAM.
3:09:50 PM      comment []  trackback []




© Copyright 2005 John Sequeira.
Last update: 5/26/2005; 9:51:17 PM.