I'll also note that the bug that killed round 1 of 1.1.1 was not found by any test we currently have. All it would have taken is a test that did any map call followed by almost any other bit of javascript (and sm 1.7.0).
On 20 October 2011 21:22, Paul Davis <[email protected]> wrote: > On Thu, Oct 20, 2011 at 1:45 PM, Randall Leeds <[email protected]> > wrote: >> On Thu, Oct 20, 2011 at 13:42, Paul Davis <[email protected]>wrote: >> >>> On Thu, Oct 20, 2011 at 1:38 AM, Randall Leeds <[email protected]> wrote: >>> > On Wed, Oct 19, 2011 at 23:38, J. Lee Coltrane < >>> [email protected]>wrote: >>> > >>> >> >>> >> > For what it's worth, a CLI based test system is what I was imagining >>> >> > as well. Take Futon out of the mix and test CouchDB. >>> >> >>> >> IMO, If CouchDB is intended to be a server that can be accessed from >>> >> the browser directly, then there should continue to be some kind of >>> >> browser-based test suite that would serve to confirm this capability. >>> >> >>> >> >>> >> I have been looking closely at the Futon tests in 1.1.0 for the last >>> >> several days, with the idea that I might begin to clean them up a bit >>> >> as time permits. >>> >> >>> >> I have found that, while some of these test failures are totally bogus, >>> >> *some* of them actually do stem from real issues -- minor >>> >> incompatibilities between CouchDB's http interface, and the internal >>> >> mechanisms of modern browsers (XHR, caching, etc). >>> >> >>> >> These are problems that we're not going to catch with a stateless, >>> >> cache-less http client running on the CLI. (I can provide examples) >>> >> >>> >> These issues have the potential to cause real problems for >>> >> developers of real browser-based apps "in the wild". That means, >>> >> there's valuable info to be gathered from the browser tests, Iff we >>> >> can clean them up, and make them behave consistently; so that >>> >> when they fail or succeed, we can actually trust the results. >>> >> >>> >> >>> >> After digging around a good bit, I can see no reason why the existing >>> >> tests couldn't be cleaned up and made to work correctly in all current >>> >> versions of major browsers. I also see no reason why the same tests >>> >> couldn't be used successfully from the CLI and `make check` as well. >>> >> >>> >> I do see significant benefits to using the same javascript test code in >>> >> all environments we test. >>> >> >>> >> -Lee >>> >> (irc: coltr) >>> >> >>> > >>> > +1 >>> > Verify Installation could grow into a suite of browser/futon tests that >>> > verify that futon (and apps in general) work, including interactions with >>> > proxies and the like. >>> >>> Sure. Client tests that test the client are fine. >>> >>> > The test suite for developers should run cleanly from the CLI as part of >>> > make check, but continue to be exposed in futon. We should work to be >>> sure >>> > they function as well as possible, for the reasons you provide. >>> > >>> >>> Blargh no. Server tests should be testing the server. The entire point >>> of moving to the command line is so that we don't have to maintain the >>> Futon test suite. Just look at the 1.1.1 thread (or damn near any >>> release thread) and the wildly varying reports of test output. The >>> situation is just a waste of time for everyone involved. >>> >>> > I think the JS testing situation is a great place for people to jump in >>> and >>> > help out, especially with the browser environment diversity. >>> > >>> >>> Sure, but I don't see what this has to do with browsers. >>> >> >> People who aren't into the internals can help to fix the suite to work in >> different browser environments. That's all I meant. >> > > Seeing as I'm having a Negative Nancy day, I'll just ask rhetorically, > "If these people exist, why do I not see anything in JIRA?" > >> I suggested that the CLI tests be exposed in Futon because I think there are >> probably some JS heads in this community who wouldn't have too much trouble >> fixing a lot of the user agent related issues in the test suite. I didn't >> mean to suggest that it should continue to be part of the release procedure >> (it shouldn't) or that we should feel 100% obligated to make sure they pass >> in 100% of environments (we can't and shouldn't), but J. Lee's point about >> how keeping such tests around can sometimes expose interesting problems we >> wouldn't otherwise see, possible outside the CouchDB codebase even, is >> worthwhile. >> >> -Randall >> > > We've had these tests for three years or more now. Perhaps I'm just > being dense today but I can't think of a single specific case where > testing things in the browser has lead to a bug report/fix that we > wouldn't have found with pure CLI tests. > > The only thing that I'm aware that the tests have done for us is > required us to exert a nontrivial amount of effort to keep them > running in multiple browser environments. I have no interest in > maintaing these as tests runnable in the browser. I want to create a > CLI test environment that promotes stable, repeatable, concise tests. > Running these in a browser is the antithesis to such an environment. >
