ShowMeDo Blogs the World » Kyran’s thought for the day

Relaunch, Refresh, Renew

wizzy — Sun, 23 May 2010 00:35:59 +0000

Not sure what it is, but it starts with re. Inspired somewhat by a good friend (thanks Steve!) I galvanised a few galvons and decided to redo Showmedo. Chief among the aims was:

Make navigation to the video-tutorials much easier/more intuitive.
Make the site less visually er weird/fussy/cluttered
Reorganize and refactor the codebase and remove the huge amounts of cruft behind the scenes to…
Free time in the future.
Make things faster. Page-loads, server-load, the works.
Giving the club a home.

I think pretty substantial inroads have been made with regard to these goals. Not perfect by any means, but there is now a reasonably solid base on which to build. I’ll be documenting the changes a bit in the next week or so, but hope the new site is a much nicer prospect.

Please feel free to direct any feedback here.

cheers,
Kyran

Python to the Rescue

wizzy — Thu, 25 Feb 2010 14:59:26 +0000

Showmedo caught some flak from the recent Westhost (our main server-provider) troubles. Catch this pretty interesting thread for the gory details. We were actually pretty lucky as the main site remained standing (not the case for many others) but did manage to lose our cron daemon. As a result some restart scripts which check on the health of, among other things, the Python Turbogears process running the site died. Now this script isn’t needed too often but if the site falls down and we’re literally asleep then it avoids more than a minute or sos downtime. The crontab line in question looks like this:

*/1 * * * * source /.bashrc && /usr/local/bin/python2.5 /home/showmedo/showmedo/cron_start_showmedo.py > /var/log/cron_log

This line tells cron to run cron_start_showmedo.py every minute and the python process will check the health of Showmedo’s main process, our memcache and mysql servers and a few other things and restart them if anything has fallen over. Having lost cron, I needed a quick hack to perform these duties and found some very cool advice here at Slashdot. For various reasons the recommended bash-scripts seemed to be dying on me, and lacking the time or will to try debugging on fairly foreign turf, I hacked a little Python substitute to do much the same:

  1 import time
  2 import subprocess
  3
  4 while 1:
  5     subprocess.Popen(['/usr/local/bin/python2.5', '/home/showmedo/showmedo/cron_start_showmedo.py'])
  6     time.sleep(60)

Not pretty, but at five lines quick and succinct. The line 4 while loop sets things running in perpetuity, line 5 uses Python’s newish (and pretty powerful) subprocess module call the cron_start_showmedo.py module and line 6 takes a 60 second break before continuing the loop.

The script sat duty while I waited for our cron-daemon to return and I slept sounder in my bed as a result.

The font of knowledge

wizzy — Sun, 17 Jan 2010 22:51:09 +0000

One of the nice things about the web and coding in general is the occasional little distraction thrown up in the space of toolsets, coding-practice and the like. Today those eddies in cyberspace have brought fonts to the fore, after a while in which I really hadn’t given it too much thought. But the fact is, for someone hacking code all day a good font, while maybe not as critical as a good editor (VIM btw), is important. In the same way as buying a good bed is important, you’ll have been staring at this font for a good few life years, when time comes to cash in your programming chips. So is worth some consideration.

I generally go for a good, reliable mono-space default which is less likely to go horribly wrong (VIM sometime gets confused with widths) and is cross-platform and easy. Andale Mono tends to do the job, without feeling particularly classy. Anyway, the Consolas font crops up a lot in various programming contexts, e.g. Slashdot, Reddit, Stackoverflow.

So my curiosity has finally been piqued and I’m going to give it a try. I certainly don’t have the balls to try a proportional font, even if readability is improved by 18% or so. Here is today’s Slashdot thread, that undoubtedly gets to the heart of the issue. And a nice post from Hive-Logic about the top ten programming fonts, which suggests I should give Inconsolata a go. Which I probably will. The accompanying Reddit thread, that undoubtedly gets to the heart of the issue.

I’ll buzz back with an opinion when I’ve had time to absorb the changes. Exciting times…

Oh, and for linux bods such as myself, here’s a little hack to get them into VIM.

Taming Flowplayer Pt. 1

wizzy — Sun, 17 Jan 2010 07:43:14 +0000

Getting Flowplayer up and running on the site was not quite as smooth as the nice web-site and clean API suggested. I’ll be documenting the whole gory process in the ‘Building a Website (with Python, jquery etc..)’ series, but in the meantime I’ll mention the hideous hackery needed to get over the final hurdle that is Internet Explorer 7.

Somewhere along the line I was looking to pass a web-address, as part of an html overlay, into the flash-player. This has the standard form:

"http://showmedo.com/videotutorials/video?name=1470110&fromSeriesID=147"

Now if one tries to pass that string into a flashplayer, all hell will break loose. Some of those symbols are going to be interpreted as controls rather than nice, neutral characters. For this reason they need to be ‘escaped’, and Turbogears provides a nice little function “quote_plus”, via Python’s urllib, to do just this. The result is this rather scary looking string:

"http%3A%2F%2Fshowmedo.com%2Fvideotutorials%2Fvideo%3Fname%3D1470110%26fromSeriesID%3D147″

This string makes it through the flash-player in all the browsers we’re interested in at Showmedo and produces a clickable flash-overlay which, strangely enough, takes one through to the page in question. Er, except (and twas ever thus), IE7 (not 8 ), where, when all the escape-artistry has taken place, we find ourselves directed to this non-existent page:

"http://showmedo.com/http://showmedo.com/videotutorials/video?name=1470110&fromSeriesID=147"

which is a very poorly URL indeed.

Now I’m guessing that somewhere at the heart of this is the issue of absolute and relative web-addresses. The second ‘http://…’ should be interpreted as an absolute address, but those unicode escape-characters have confused the system, making it treat the URL as relative, thus adding another root-url.

If I had time to explore this horror I would probably spend it doing something else, but in these time-starved times I just did a little bit of playing around and added a hack-function which undoes the ‘:’ and ‘/’ escapes of “quote_plus”, producing this:

"http://showmedo.com/videotutorials/video%3Fname%3D1470110%26fromSeriesID%3D147″

Which makes it through the browser minefield, but, as seems so often the case with this messed up web-development world, at the cost of some serious inelegance.

New flash-player being debuted

wizzy — Fri, 08 Jan 2010 22:24:21 +0000

It’s been much more work than anticipated (a definite screencast or three in there somewhere) but we’ve finally moved to the open-source flowplayer Chief benefits are:

much cleaner jquery-like API
far greater control over the video-player area. Which means…
innovative ways of directing users to new content and…
with luck much more interactivity with the video-player

p.s. the commenting on-site is getting a much-needed overhaul. With a bit of cookie-magic we should be able to get some dialogues going and vanquish those nasty anonymous comments (mea culpa).

A Wing, a Prayer, and the Ol’ Big G

wizzy — Sun, 22 Mar 2009 02:47:11 +0000

Fresh from bemoaning increasing Google intrusiveness , here’s a little tale to warm the cockles. Showmedo’s email wented AWOL today, cue five hours of so of adrenaline-fueled panic. A few hundred forum-trawls later, it turns out that our server-host has a bit of previous on the old email-front, as a stereotypical UK policeman might observe (lawks but writing for an international audiance doesn’t half give one pause missus). Well using my google-search-fu and some Python-based smtp-lib hacery I was finally able to bypass our ailing local webserver and use the rather convenient and pretty robust Gmail-based SMTP server facility. It all feels wrong and hacky but, gosh darn it, it works and might prove to be one of those long-term short-term fixes.

The downsides may well make themselves known in the morn’ but right now I’m too knackered to care. Feel free to test our email by posting some very warm comments to your favorite author.

A Sting in the Tale

wizzy — Mon, 16 Mar 2009 12:47:04 +0000

I just caught a very cool example of what some people are doing with javascript these days. A little molecule-builder, it’s impressive and fun at the same time:

check out Molecools

Now that was fun, and note the example molecule, Ethanol. As an aside, the simulator is pretty crude so there’s no valency-counting or any of the refined stuff that lends molecules their particular shape and characteristics but, like all javascript, the code is out there in the public domain to peruse – just give the author an acknowledgement if you use anything substantive.

Anyway, as I noted, the molecule demonstrated was Ethanol, which is a nice, definite unambiguous noun and has not formed any part of my browsing history for a while or so. So when I returned to my Gmail account and saw the little top-advert for some Ethanol related products, there was no doubting the provenance.

Now Google’s permeation of my browser and their improved ad-targeting has been an expanding wedge for a while and I always assumed they were making note of my google-searchs and using them in my gmail-ad, although I’d never been able to unequivocally pin them down here – I have a lot of emails for them to farm data from. But here was something a bit different, unequivocal evidence of them using a link-click (from reddit as it happens), unrelated to any google-search, as the basis for a targeted ad. I managed to find this little explanation of what appears to be a new phenomenon:

http://www.dslreports.com/forum/r22052924-Google-Uses-Your-Browser-History-To-Target-Ads

I’m assuming reddit is part of some affiliation scheme. Now I don’t know about you but this has crossed my Rubicon as far as Google intrusiveness is concerned, and their opt-out scheme, as explained in the article is a bad one. I think I’m going to hunt out one of those Firefox gmail ad-blockers; to be honest the sheer banality of most of the Forbes promos or jokey comments for the day was tending me that way. Anyway, another reason to keep an eye on the big G, our self-appointed (and generally benign?) dictator.

No related posts.