Posts tagged: WEB DEVELOPMENT

Think twice before you let those domains lapse

In an effort to shed time-sucking side projects in the past couple years I’ve let a number of sites go dark and domain names lapse. Some of these were ideas that never got off the ground, but one or two were sites with real traffic and Google pagerank (PR 5 in one case, not stellar but not achievable overnight either).

Sadly, some of these domains have now been taken over by those useless squatter pages that manufacture lists of “related links” and “popular searches” and so on to trick people into clicking on ads. This only makes sense – if you were a domain squatter, you’d certainly prioritize grabbing expired domains with high pagerank and many existing inbound links. If you set up your server never to return a 404, some of those linking sites might never even notice the change.

SourceForge needs help

Despite their adoption of Big Green Download Button technology, SourceForge still has an absurdly cumbersome download process. I know it’s annoying to just gripe (I try to see the positive side too); I’m just surprised that it’s still this crufty. According to the OSTG site, “SourceForge.net hosts more Open Source development products than any other site or network worldwide.” But if the pace of modernization doesn’t pick up, I’m afraid that won’t be true for much longer.

Root, sweet root

For several weeks at work I’ve been prepping for a server move; this week we flipped the switch. It was the most serious migration I’ve ever done, and it went very well. Some notes:

  • Previously we shared a dedicated box at a certain very large colo provider. A few weeks ago, in the course of swapping out a failing drive in our box, staffers at the colo 1) wiped out the live backup of the drive and then 2) destroyed the contents of the failing drive, overwriting it with a week-old backup. A week is a lot for us, with a couple limited-access applications (Django apps, naturally) seeing steady daily use by hundreds of users. This just firmed my resolve to go with a smaller provider (JohnCompanies) who has taken good care of me over the past three years.

  • The other thing the above incident drove home is that backups are good. We lost no data in the end, but recovering some of it was a real pain (in one case reconstructing database inserts from logfiles). So, the new server, in addition to being backed up daily onsite, backs up daily to an rsync.net server. There are also hourly and daily cronjobs making database dumps – at any given point I have a snapshot for each of the previous 24 hours and one for each of the previous 7 days. And finally, we can pull down local copies from the rsync.net server. It’s about a gig of data right now; I burned it to two CDs and brought them home for safekeeping.

  • There are three Django projects running on the server, and more to come. All back-end content administration is handled by Django, and in the fall I plan to replace the remaining front-end PHP pieces. Having full control over the box and Apache makes my Django work much easier. Perhaps oddly, I have three different versions of the source running in parallel (0.91, pre-merge magic-removal, and current trunk), roughly reflecting the times when the three projects were created. These are all straight Subversion checkouts living in /usr/local/www/django, and I set the PYTHONPATH appropriately for each in my Apache config.

  • I had tested most of our software configuration on a play server, so setup was fairly straightforward. The glitches were minor and familiar: an “HTTP Accept” filter warning on Apache startup, a thread-related workaround for Apache, and a stable SVN checkout of mod_python (rather than the current 3.2.8 release) needed to work with Apache 2.2.

  • I’m using portsnap instead of CVSup, and ssmtp instead of sendmail.

  • The fact that our email is handled by a separate provider (not my department, thankfully) made this much easier than it otherwise would have been.

  • The shared colo box was well run – but there’s no substitute for having root!

Trying to send eBay a message?

I’ve been getting unrequested messages from the eBay Developer Zone site over the last couple days about email address changes. I have a Developer Zone account, but I haven’t touched it for months if not longer. I sent an e-mail to them about it, but haven’t heard anything.

I just got yet another one of these and I think I see what’s up. Check out the registered email address. Somebody found a hole.

Web developer evolution

After the frameworks post I kept thinking about this. Of course any generalizations I make are heavily colored by my own direct experience, but the progression seems to go along these lines:

  1. Make static web pages.

  2. Make modular pages using simple includes (in SSI, PHP, ASP, or what have you).

  3. Make pages with more involved functionality (form submissions to database, basic CRUD).

  4. Get sick of re-building dynamic stuff for every project; write your own kinda portable library or framework.

  5. Get sick of maintaining your own framework, then:

  6. (a) find an existing framework that fits your philosophy and has support (commercial or open source community); use that. – OR – (b) Start an open source project to build the ultimate dream framework, then:

  7. (a) Release it to great acclaim – OR – (b) Watch it wallow in obscurity; return to step 6.