60 pages of discussion and @DigitalScience throwing the first stone.
It’s been a good couple of years since I had a poke around here, good to know some things never change
Thanks so much for all the well wishes. I’m feeling very warm and fuzzy on the inside.
I’ll definitely be around the marketplaces still in some way. I suck at design for a start, so will probably be cruising back for templates & designs at the very least.
Plus I’m sure I’ll come troll the odd forum thread or two.
I’m sad and also happy/excited to say I’m moving on from Envato – see the notes post http://notes.envato.com/developers/goodbye-cruel-world/
It’s been a great 4 years hanging with you all, I’m gonna miss working with such an awesome user community.
Don’t know if it’s related, but Statementer is broken for me (it displays some info but shows “Invalid date” everywhere) http://gyazo.com/ce507d0cbd874646a2015899c56b811b http://gyazo.com/7041306aa5b75b4e6c6c64105cf404cf http://gyazo.com/3255cf58586497bbe2c59e23408a1fcc
Yep, that’s the aforementioned API date problem. We’re working on it.
There are some API bugs hanging around from yesterday’s ruby upgrade – see http://activeden.net/forums/thread/ruby-upgrade-around-noon-16212/59473?page=12#533332 for details
That was probably the least pleasant 24 hours the whole team has been through in quite a while.
We tracked down the root cause of the intermittent blank error pages – the way our web server handles unicode in urls could be described as “very poor, would not buy again”. For those who might be interested, you can see the test case Lucas made for us to submit to the group who maintain it over at https://github.com/lparry/how_to_kill_a_unicorn
There are still some small problems left that we’re working on:
- A certain percentage of our photo processing servers won’t boot the new code (for no discernible reason) so there might be some delays in photo processing
- Date formats are still different in the API , once we find a neat and tidy way of returning them to the older format we’ll do so
- some of our internal reporting tools are broken
The search bug is unrelated, same with the comments one, but they’ll get looked at once we lock down the ones above.
Please keep reporting any oddities you find around the place, and thanks so much for being so supportive through a very tricky upgrade.
OK, so while we “fixed” it a couple of hours ago it looks like it came back. I just restarted all the web servers and it looks OK.
I now have a new working theory: with Ruby being a dynamic language, the whole site isn’t actually loaded into memory when we start the web server, but bit by bit as the site gets visited. I think that something is still a little broken on a very rarely visited page… so after a couple of hours someone visits the bad page and it poisons the web server.
I’ve not had any luck finding the code right now, but given it looks like the site stays good for a while after a refresh… I’m gonna sleep a little longer then head to the office (with my much more powerful computer and ready supply of coffee) to sort it once and for all.