Tests for VE hidden in our browser tests show this is broken again.
This prompts 2 things: 1) We are still not doing a good enough job of ensuring that VE keeps compatibility with mobile. James - is there anyway we can alert you better about these issues, maybe running our browser tests for VE alongside yours? 2) We ___really___ need to get our browser tests into better shape. This could have been catched sooner had Chris not managed to wade through the tests to actually assess the false positives.
Let's try and get this fixed asap so broken code does not get on the deployment train...
On the plus side, our tests are working (sometimes :-))
On Tue, Mar 25, 2014 at 11:21 AM, Jon Robson jdlrobson@gmail.com wrote:
- We ___really___ need to get our browser tests into better shape.
This could have been catched sooner had Chris not managed to wade through the tests to actually assess the false positives.
Actually, this failure stuck out like a sore thumb after breaking overnight, and there was only a little extra noise from another test. No additional overhead.
I do want to make the MF tests better though. I am planning to be All MobileFrontend All The Time for about two weeks starting Apr 2. (I am trying to clear the decks of VE and QA Lead stuff this week to free up as much time as possible.) -C
On 25 March 2014 11:21, Jon Robson jdlrobson@gmail.com wrote:
Tests for VE hidden in our browser tests show this is broken again.
In future, please e-mail ve-tech like before so this gets found.
This prompts 2 things:
- We are still not doing a good enough job of ensuring that VE keeps
compatibility with mobile. James - is there anyway we can alert you better about these issues, maybe running our browser tests for VE alongside yours?
The browser tests *do* run alongside each other — every 12 hours. They told us exactly that something had broken in the last 12 hours of code merges, and I was planning to look at them today to work out if it was just the usual. Pinging us in IRC would have helped us fix it faster, but < 2 business hours is OK, if not great sped for fixing.
Let's try and get this fixed asap so broken code does not get on the deployment train...
Already fixed. But yes.
J.
On Tue, Mar 25, 2014 at 12:14 PM, James Forrester jforrester@wikimedia.org wrote:
On 25 March 2014 11:21, Jon Robson jdlrobson@gmail.com wrote:
Tests for VE hidden in our browser tests show this is broken again.
In future, please e-mail ve-tech like before so this gets found.
Noted.
This prompts 2 things:
- We are still not doing a good enough job of ensuring that VE keeps
compatibility with mobile. James - is there anyway we can alert you better about these issues, maybe running our browser tests for VE alongside yours?
The browser tests do run alongside each other -- every 12 hours. They told us exactly that something had broken in the last 12 hours of code merges, and I was planning to look at them today to work out if it was just the usual. Pinging us in IRC would have helped us fix it faster, but < 2 business hours is OK, if not great sped for fixing.
Thanks for this. I wasn't aware of this but this is good. I just saw the patch raised against MobileFrontend and wasn't sure if you guys were aware of it. This wasn't meant as an attack of any sorts I just wanted to explore if there were ways we could improve our workflow to prevent huge breakages like this - my worry is had this change been merged on a Wednesday we would have had to do a lightning deploy or had VE broken on all wikis for a week which would have sucked.
Let's try and get this fixed asap so broken code does not get on the deployment train...
Already fixed. But yes.
Thanks for the speedy fix :)
On 25 March 2014 14:07, Jon Robson jdlrobson@gmail.com wrote:
On Tue, Mar 25, 2014 at 12:14 PM, James Forrester jforrester@wikimedia.org wrote:
On 25 March 2014 11:21, Jon Robson jdlrobson@gmail.com wrote:
- We are still not doing a good enough job of ensuring that VE keeps
compatibility with mobile. James - is there anyway we can alert you better about these issues, maybe running our browser tests for VE alongside yours?
The browser tests do run alongside each other -- every 12 hours. They
told us
exactly that something had broken in the last 12 hours of code merges,
and I
was planning to look at them today to work out if it was just the usual. Pinging us in IRC would have helped us fix it faster, but < 2 business
hours
is OK, if not great sped for fixing.
Thanks for this. I wasn't aware of this but this is good. I just saw the patch raised against MobileFrontend and wasn't sure if you guys were aware of it. This wasn't meant as an attack of any sorts I just wanted to explore if there were ways we could improve our workflow to prevent huge breakages like this - my worry is had this change been merged on a Wednesday we would have had to do a lightning deploy or had VE broken on all wikis for a week which would have sucked.
Oh, completely. It's why we try not to merge big things on Wednesdays.
J.