USPS goes open-source with tracking system (GCN) [LWN.net]

USPS goes open-source with tracking system (GCN)

Posted Jul 10, 2009 20:31 UTC (Fri) by jspaleta (subscriber, #50639)
In reply to: USPS goes open-source with tracking system (GCN) by corbet
Parent article: USPS goes open-source with tracking system (GCN)

You bring up an interesting point. Assuming you dropped everything you were doing and did the absolute minimum necessary to open the codebase..would it be worth doing? Would it be likely that a community would grow up around it?

Could you take the time and use spot's scoring system and tell us where the LWN opening effort would fall on the failure scale?

http://spot.livejournal.com/308370.html

While spot's list is both humorous and angry (and more humorous because of the angst) he makes a very valid point. Throwing code over the wall doesn't make the code instantly valuable as a piece of the open source ecosystem that can be nurtured and sustained. It does take effort to prepared for success in terms of building a community development model.

For me the question isn't why isn't the existing LWN codebase not open. I accept that the code you are dealing with is high maintenance, custom crap that you aren't particularly proud of nor want to show off but gets the job done. I am myself a master of such..artful..plumbing work. Both in software and in real life. Everyone is happier if they never watch how I sweat a copper pipe or write php code. And not just because I tend to do both in just my underwear.

The question is, why aren't you using an AGPL open solution that will in the long term lower your own maintenance burden by allowing you to share that burden with a community of users/developers who share pieces of a larger solution?

-jef

(Log in to post comments)

LWN site code

Posted Jul 10, 2009 20:47 UTC (Fri) by corbet (editor, #1) [Link]

Could you take the time and use spot's scoring system and tell us where the LWN opening effort would fall on the failure scale?

It would be up there, but probably unfairly so. We don't have things like "releases" or build documentation because there's only one deployed site. For similar reasons we have no mailing list, no web site, etc.

It's actually not all that bad of a code base, considering; I'm not ashamed of it. But, naturally enough, neither is it particularly polished or ready for more general use.

The question is, why aren't you using an AGPL open solution that will in the long term lower your own maintenance burden by allowing you to share that burden with a community of users/developers who share pieces of a larger solution?

If we were starting now we'd probably do that; there's a lot of good stuff out there. When I did start, back in 2002, I based the first version on what seemed to be the best option available at that time: Zope/CMF. But then I got distracted for three months, came back, and realized that I had absolutely no clue of what my code was doing. That was a shock - Python isn't normally like that. So I decided I was better off doing something new, and I've not regretted that.

Now there's so many good options that it would make little sense to start from the beginning. We've talked occasionally about switching to something more mainstream, but that would be a world of pain too.

USPS goes open-source with tracking system (GCN)

Posted Jul 10, 2009 20:56 UTC (Fri) by dlang (✭ supporter ✭, #313) [Link]

exactly what AGPL codebase do you suggest? while it will implement lots of things that LWN doesn't currently have, does it implement everything that LWN _does_ currently have? (including the subscriber pass-along links?)

how much work would it be to migrate all the content from the current system to the new system?

how big a problem is it that all the persistant links to the current content would no longer work with the new system? (remember that some of these links are in kernel commit messages, you can't change them, even if you could somehow track them all down)

frankly, I like the way LWN works a lot better than just about any other site I deal with.

it's simple, clean, and to the point.

all of the 'popular' systems that I have seen are written by people who would instead use terms like 'pathetic', 'obsolete', possibly 'boring' if they were being kind.

I would like to have the code for the LWN site available, but not at the cost of taking enough of Jon's time that LWN suffers as a result.