drewp

I want something passively uploading my GPS data. What should I get?

I don't see how latency would be a big problem for file storage. Home user file reads and writes are very spatially coherent, right? Keep the first N bytes of each file in a local cache, and when the user starts to read a file, you're pre-buffered. Note that this is just a cache; the whole file is still reliably stored at the remote site.

Users probably use the same small subset of files over and over again, too, so you don't even need to keep the head of all the files to eliminate most latency effects.

That was a great way to handle the subject line (original title plus 16-word summary). Thanks!

Why don't you pair it? The advance with opencalais is that they offer a production-level service that's very easy to use and integrate. (Especially when they get normalization working.)

Jump start != produce a complete end-to-end solution including whatever components you happen to think are essential

I agree. Every program seemed to have a few "while (true)" lines, often with comments above them like "// infinite time loop" :(

And another example for your earlier point:

// print out the shred id <<< me.id(); >>>;

There's a nice demo of the classifier API at about 40 minutes into this video: http://www.youtube.com/watch?v=keXW_5-llD0

I will. That page might have been a nice intro to the field, but there's a much smaller version of it that I'm going to remember and that I'm going to tell other people. That's the elevator pitch.

It also doesn't help that the "what we do" page starts by criticizing people who call them a data mining company. They apparently do make software for use in (one definition of) data mining.

The proof of Proebsting's Law seems kind of silly because running a compiler with optimizations turned off wouldn't actually disable all the years of advancements of compiler technology. It seems quite reasonable that many compiler advancements have to be added at the same time as hardware support. That doesn't make them purely hardware contributions.

Also, the proof skips more than a few critical steps when it says "Let's assume that this ratio is about 4X" and then uses that number as a multiplier in the answer.

Why is this so emotional for you? Writers barely know if you're reading, so it doesn't seem worth any stress or "overwhelming" feelings for you to work so hard on selecting feeds or catching up on them.

Where are we, on digg.com?! The subject line of this post should have said "eat less" in it. I shouldn't have to go to the article to get those two words.

Listening to http://www.sky.fm/soundtracks/, perhaps. Their library isn't huge, but it's still fun.

Don't forget the costs of using a "lightweight" markup:

You have to learn some new system of quoting and escaping. I don't know textile or markdown, but I just went over to http://textism.com/tools/textile/index.php and tried to put a </code> close tag inside the sample code. It's got problems.

You get approximately one chance at a correct conversion to a richer format: converting to html. Nearly any other conversion, such as from the wrong lightweight markup to another lightweight markup, is lossy.

These markups aren't very extensible. They can't store metadata (e.g. microformats), or if they do, it tends to be by allowing html tags mixed into the markup, which means we're back into quoting/escaping issues.

There tend not to be any WYSIWYG editors, neither for you if you one day want one, nor for a less technical user who you'll someday want to edit your document.

So given almost zero knowledge about your problem, I'd recommend a WYSIWYG html editor. Whichever one you choose, it'll occasionally screw up and corrupt your formatting. Or maybe there will be subtle differences between the preview layout and a final browser layout. If you think those kind of issues will be a problem, then you REALLY don't want a lightweight markup language.

http://freebase.com would probably like to gather and hold your data. They have a nice GUI for editing it. People are starting to make other sites that use freebase, e.g. http://www.dipity.com/

http://youtube.com/results?search_query=artoolkit&search... for some neat demos of OSS software to overlay computer graphics on real targets

http://thecoccinella.org/ "Coccinella is a free and open-source cross-platform chat client with a built-in whiteboard for improved collaboration with other people. "

(uses jabber)

Bug tracking shouldn't be hosted, in general. I should be able to declare a bug in anything by posting it on my blog, and people can do their commenting/triage/fileposting/whatever else they do by using Regular Internet Technology. There is a data format standard issue to be solved and a few search issues.

A project can have a central bug tracking site for people who don't have blogs, but if I don't like that UI I should be able to switch to another.

My proposal also helps a lot when issue tracking is spread across companies. I find an in-house issue which I report to my own team; and then they want to forward it to some outside vendor. Much effort is wasted maintaining the links between the system used by the in-house team and the one used by the vendor.

http://code.google.com/p/baetle/ has some plans for the data model and format.

Way to scare 'em back to using bottled water again, AP! Yes I know bottled may not be any better, but it will certainly feel safer to people who want to Just Do Something in response to that FUD article.

position: absolute vs relative

I've had to look up the rules several times now for how the parent box is chosen. I still don't know if I'm making more intermediate divs than I have to.

I apologize for the slang use of the word 'hacked' up there.

I meant 'vandalized, probably by means that required deep knowledge of a system to initially develop, although the vandal was unlikely the same person as the hacker who came up with the exploit'.

His site was indeed hacked recently, and it was harming my computer inasmuch as it was crashing firefox. I consider the google warning to be late but not incorrect.

Here's an email I sent to Joe on 2007-12-17. (It's really hard to make these kinds of messages not look like spam!)

----- Hey, I was checking out iui tonight, and your site seems broken.

The front page has JS that does something involving "fredkidns.com" and it repeatedly causes my firefox to fill memory. Kazehakase is able to show the broken page, though.

I think the blog top page is also hacked; other internal pages seem to be ok.

This post invites a ridiculous "me too" thread.

I think you meant to write "if you read hacker news, please bookmark http://news.ycombinator.com/ if you haven't done so already".

Then we can have a real discussion in the comments section, you get the usernames listed in delicious in a way you can click-to-browse them, and you even get a few more hits from people who care about hacker news but didn't think your particular post was worthwhile :)

Python totally has more than one way to do it. People sometimes simplify the python approach as 'the opposite of the perl approach', but they're not exactly opposites.

The philosophy that makes python distinct is "There should be one-- and preferably only one --obvious way to do it." Emphasize the word 'obvious' if you don't see why this isn't the same as "There is only one way to do it".

Python has many redundant mechanisms: def/lambda, list.append/list.extend, sys.exit/SystemExit, x[y]/getattr(x,y), etc. But even though multiple mechanisms exist, the hope is that there will be a single obvious way to do whatever you want at the moment. Want to add one element to a list? list.append. Want to add another sequence to a list? list.extend. (The point here is that list.extend would be technically sufficient for both cases, but it makes the one-element case clunkier and less obvious.)

A language where there was actually only one way to get a particular result might be interesting, but it wouldn't look like python.

newsbruiser:

pretty friendly install and web config

can't avoid running it as cgi, which sucks

once I decided to make my entries in restructured text (huge mistake), I can't easily change to using regular html for new ones, which is what I want. Half my entries include an apology for not being able to present some fancy escaping correctly or for screwing up my links. I just want to use xinha! This is likely the issue that will finally cause me to switch blog systems.

spam comments have corrupted stories in the past. A look at the gross NIH article file format will show you that comments-corrupting-stories would not be surprising

anti-spam comment system is too hard to get right. I'm currently on the side of "reject all good comments" and I don't know how to get to a better state

it has tags, but I can't provide an RSS feed for a single tag, which I'd really like

it's in python, which is why I chose it, but the barrier to writing patches or plugins (e.g. per-tag rss feeds, or per-article input syntaxes) is too high

each blog gets a large number of scattered data files, including some dotfiles (why?!), which can make it hard to reorganize blogs or get the perms right when you move everything to a new NFS drive

Thanks for asking!

The advice wasn't "always choose the harder thing". Obviously that's not correct, even generally speaking.

Note the part where you're supposed to do this _when you have a hard time deciding_.

There would be almost nothing in common with your non-FB "toolkit" and the FB one. Theirs is a way to access the FB database, yours wouldn't be.

Hopefully you're not imagining that FB for some reason decided to give away the tools that would be needed to create another FB. That's not what they mean by "toolkit".