jcrocholl

  base64: bThvW3JKYz16bXpUWDIufkl2OllseC4maV1gMjtVYXNtb2RyZ3czbkQzQDtnZnRdMVZuZkIqVTJpIGlxa2xubkAoaTRueGdQaGN0YyRjTGN1WGlUbzF5VU5qcz50KmRwZXlncGN0dA==
  text: m8o[rJc=zmzTX2.~Iv:Ylx.&i]`2;Uasmodrgw3nD3@;gft]1VnfB*U2i iqklnn@(i4nxgPhctc$cLcuXiTo1yUNjs>t*dpeygpctt
  letters:  $&(48:=>BDIJLNPVY[`aehjkquvw~*.13;@TX]dflprsxyz2Umogintc
  counted: 111111111111111111111111111111222222222222222222333345556
  progressively hiding infrequent letters:
  m o r c zmzTX2.     lx. i] 2;U smodrg 3n 3@;gft]1 nf *U2i i  lnn@ i nxg  ctc c c XiTo1yU  s t*dp ygpctt
  m o   c  m   2          i  2 U  mo  g  n    g t   n   U2i i   nn  i n g  ctc c c  i o  U    t     g ctt
        c                 i           g  n    g t   n     i i   nn  i n g  ctc c c  i         t     g ctt
        c                 i              n      t   n     i i   nn  i n    ctc c c  i         t       ctt
        c                                                                  c c c c                    c

Clickable link: http://boardlite.com/

I like the look, it is clean and simple.

One small glitch: it seems that on the page http://boardlite.com/category/computers/list/ some vertical gray border is missing on the left side of "1 post" and "16 views", where the light blue headline becomes narrower and says "Tracking ID: 400000001" etc. Same problem on the FAQ page at http://boardlite.com/faq/ ...

Page load speed: instead of http://boardlite.com/media/js/lib/jquery/jquery-1.3.2.min.js you could use http://ajax.googleapis.com/ajax/libs/jquery/1.3.2/jquery.min..., see http://code.google.com/apis/ajaxlibs/ for details.

Point taken. I have now integrated the display of taken (and parked and error) top level domains again, trying to make it useful but not too distracting.

Clickable link: http://www.nxdom.com/

Writing may help you more than reading, especially if you can find somebody to give feedback. Maybe start a blog or contribute to a magazine?

I'm not a native English speaker, but here's some feedback for you: I would say "advice" (always singular) rather than "advices".

I call myself "Founder" in the context of my current solo project, even if that project is not really a "startup" or "company".

As the legal term I use "Sole Proprietor": http://en.wikipedia.org/wiki/Sole_proprietorship

Thank you, I see what you mean. The headline wraps because the font is too wide on Ubuntu (in Firefox too), hope to fix it soon.

My interactive search engine for short available domains: http://www.nxdom.com/

I'm working to bring the advanced weighted sort order back, but optional so that new users can play with the simple sort order until they understand what it does. Work in progress is here, and full source code: http://master.latest.scoretool.appspot.com/ http://github.com/jcrocholl/nxdom

My 2-year-old son doesn't want to go to sleep yet, so he's climbing around on me while I write this.

Good question, so I wrote a blog post: http://nxdom.posterous.com/why-is-nxdom-open-source

Yes, that was the previous version that I announced here on January 1st. I figured that usability is more important than fine-grained control over the score weights, because the sort order is not terribly critical. Basically, the user should be able to adjust the trade-off between short and popular names. But it's a brainstorming tool, we're not trying to get one particular name to the very top of the list.

Because it's a search engine for short names, using a large database of unused and expired domain names. Try searching for "hack" + "ne", or "hack" + "e" for even more results.

Actually, I have now enabled the search indexes for the following prefix+suffix combinations: 4+1, 3+2, 3+1, 2+1, 2+2, 1+1, 1+2, 1+3, 2+3, 1+4. I may add longer combinations later, but for now they will be shortened and filtered, e.g. 5+3 will use cached 4+1 and then filter the results in the browser with JavaScript.

Yes, I know the combined prefix and suffix search doesn't work properly yet. The reason is that the App Engine datastore doesn't support fancy search, only sorting with custom indexes. All results for a GQL query must be adjacent in the index. The indexes for combined prefix+suffix search are still building but should be ready in a few days.

Thank you all for the great feedback!

1) You're right, but the boxes (and columns of numbers) are probably going to disappear in the next version. Maybe we'll have sliders or radio buttons to adjust between "Important" and "Don't care" for some properties.

2) Real substring searches are not supported because they are expensive. But your example is supported: type "debate" in the second keyword input field for suffix matches.

You're right. At some point I'm going to start using http://getsatisfaction.com/ or think about moderation/filtering. But for now I really like the direct feedback from users, so I'm going to delete bogus entries manually from time to time.

Congratulations to our first customer! Since my post on HN we had 612 hits on moniker.com, with 1 conversion for $0.31 commission. I also spent $1.60 today for 22.43 CPU hours on Google App Engine. Almost all of the 17k requests today were AJAX queries for domain searches.

Because we can't run DNS queries inside Google App Engine. Instead, I query the DNS results from my home computer and feed them into the datastore using remote_api.

Yes, that would be interesting because the startup time for loading the zipped Django 1.1 package is significant. Maybe app-engine-patch can be tweaked to use_library()?

I understand that's annoying, and hope to improve it. The system reports digit.com free because it doesn't have an A record, but I will fix this by checking SOA records in the future. I'm not sure why 6af.com is reported free, maybe there was a temporary DNS problem when I checked it. It should be updated in a couple of days.

My list is a fairly random sampling of the huge search space of available domain names. I'm planning to keep adding more names in the future, trying to strike a balance between good names (short but readable) and more names (requiring more storage). Fortunately, the App Engine Datastore has linear complexity for the number of query results, so adding more names doesn't slow down the searching.

Because app-engine-patch includes more than just a zipped package of Django 1.1. It comes with jQuery, the Blueprint CSS framework, a media generator to join and minify your CSS and JavaScript, automatic URLs for each app (using urlsauto.py), and helper functions like ragendja.template.render_to_response and ragendja.dbutils.get_object_or_404.

Google now supports both Python and Java on App Engine, but you're right it's Python.

It's written in about 2500 lines of Python, using Django 1.1 (app-engine-patch) on Google App Engine. Full source code is here: http://github.com/jcrocholl/scoretool

That sounds good, I'll look into it.

Yes, you can enter very negative weights in the input boxes for digits (numbers) and dashes (hyphens). Then these domains will be sorted down to the end of the list (below the end of the page).

Some of the names are made up (combinations of letters or short dictionary words). The expired names are from http://namejet.com/Pages/Downloads.aspx and I'm planning to keep adding more names in the future.

Because magic.com doesn't have an A record. If you click on the "free" link, the registrar (moniker.com for now) will tell you that it's taken. I know this is disappointing, but I haven't found a good way to do bulk whois for millions of names, DNS is much easier.

This is a long article but very useful, even if all you learn is how to use "git bisect".

Yes, but the acoustic model (or models if you account for male/female and different accents) can be improved with all the input that the service gets over time. The bad accuracy can be fixed by human intelligence using Amazon Mechanical Turk.

You mean like Twilio or lavarockhq.com? I know it's hard but I consider that an advantage. But I'm still brainstorming, exploring existing solutions, and working on a very basic demo to see if it can be done with open-source software.