#techempower-fwbm

[FrameworkBenchmarks] aschneider-techempower pushed 2 new commits to master: http://git.io/2OGk7g

FrameworkBenchmarks/master 4de6c72 Alex Schneider: Fix minor bugs with installer (mongodb should work properly now)

FrameworkBenchmarks/master fc9d013 Alex Schneider: Merge branch 'master' of https://github.com/TechEmpower/FrameworkBenchmarks

[FrameworkBenchmarks] snoyberg opened pull request #1086: Updates to Yesod and WAI (master...master) http://git.io/3VFzAw

[FrameworkBenchmarks] msmith-techempower closed pull request #788: fix(Yesod): Resolve dependencies and change executable filename in cabal... (master...master) http://git.io/mF-kUQ

[FrameworkBenchmarks] msmith-techempower pushed 6 new commits to master: http://git.io/XgPRhg

FrameworkBenchmarks/master d647df3 Michael Snoyman: Some Yesod cleanup/optimization...

FrameworkBenchmarks/master 1ce2175 Michael Snoyman: Use stylish-haskell on yesod.hs

FrameworkBenchmarks/master 47ba68e Michael Snoyman: Reduce library dependencies in yesod

:-/ According to travis about half or more of these tests don't pass. All I did was update the PHP version, but tests across all languages are failing miserably

that was happening before

if you update anything outside of the frameworks/ dir, travis runs all the tests

you can test your changes specifically by commenting out all the travis lines except that ones that interest you (just remember to uncomment those lines on your last commit to the PR)

:( In https://travis-ci.org/TechEmpower/FrameworkBenc... they are all passing

that was initiated from a #1086

pr1086

I forget what I have to say to make the bot link it

hey, the bot's gone

or not...

pr#1086

>_<

there's something I can type that'll make the bot link it

gh#1086

Updates to Yesod and WAI: https://github.com/TechEmpower/FrameworkBenchma...

there it is

anyway - that travis run was only run against wai/yesod (you can tell by the duration; anything less than 10m means that the test spun up, decided it didn't need to be run, and returned a success value; this allows us to see rather quickly whether a pull request breaks anything)

I figured something must be cached in there. Nice.

as I mentioned before, if you have files that changed beneath the frameworks dir and that's it, it will attempt to ONLY run those tests

from that one i linked, all the files edited were either under Frameworks/Wai or Frameworks/Yesod

so when travis kicked off, those were the only two that took the time to verify, all the others "succeeded"

when you eventually create your pull request, mind that if you have changed anything outside of the frameworks dir (say, in the setup dir) then ALL the tests will get kicked off on travis

this is a problem because many of them are failing and it will take 24hours to run

Yep, which is what I had to do in order to update the base version of PHP that gets run

Lots of those tests are failing out of the box anyway

yeah

the PHP tests are all failing, and I have no idea why (though, I haven't spent any time looking into php specifically)

For me, upwards of 75% of all tests in all languages fail in some capacity.

that number seems high

manual counting suggests 65/123 are passing: https://travis-ci.org/TechEmpower/FrameworkBenc...

and 24 of those are php >_<

also, these were after ashneider fixed (attempted to fix) the postgresql bugs; there is currently a problem with the install script keeping postgresql from being installed/configured correctly, so any test relying on that fails

That could be a major source of pain for the PHP tests. What's the problem with postgres install?

I did notice some issues in my work where postgres was puking because the install was assuming 9.1 instead of 9.3

Or maybe vice-versa

yeah, i'm not recalling specifically

the installation is expecting 9.3 on ubuntu14 as a minimum... i think

and travis currently only supports ubuntu12 and therefore postgresql 9.1

so there might be a problem there

I ran into that same problem when I was testing

oh actually

i'm looking at the logs on travis right now and it's using postgresql 9.3

so I think that issue is resolved, but we still have a stale build using the broken postgresql setup

The travis.yml file asks for the 9.3 add-on

ah, there might have been a workaround there

https://travis-ci.org/greg-hellings/FrameworkBe... Mine is failing all over the place even with the 9.3 requested

But not every single PHP test is failing. So I guess that's good

https://travis-ci.org/TechEmpower/FrameworkBenc... every single php test is failing

oh

checking yours

Heh, sure enough

wow, that's wacky

well, hopefully its resolved

I don't really see how to discern the problem

Oh, there, there are errors output directly to the page

^

Some of these look like straight up errors in the framework code that I might be able to address

Those within PHP, that is

[FrameworkBenchmarks] msmith-techempower closed pull request #1016: Cpoll fix (master...cpoll) http://git.io/IYjGkw