I feel like you are trying to convey the impression that finding bugs in julia results in insults and no help from us. That is a total mis-characterization of the project. There is also no equivalence between responses to bug reports, and responses to blog posts. As far as I know, all 9000 of our bug reports have been received with gratitude. However your post says or implies that we don't care about error handling, tell people to fix their own bugs, and even that we don't understand our own code. You can very well expect some pushback on that. On Dec 29, 2014 6:42 PM, "Dan Luu" <[email protected]> wrote:
> Welp, I ended up checking up on this thread again because of a > conversation with Stefan, so here are some more responses. > > I tried https://github.com/dcjones/Gadfly.jl/issues/462 on the current > release binary on julialang.org and it still fails as before, so that > wasn't because I was running off of master. I updated the issue. > > Yes, I agree that I seem to run into an unusual number of bugs. My > guess is it's partially because I basically don't do any of the kind > of data stuff people normally do with Julia and I'm off doing stuff > that's untested and has rarely, if ever, been used before. But IIRC, I > ran into a bug where code_llvm and code_native would segfault. > Sometimes stuff just breaks and doesn't get fixed for a while. > > I don't really want to get sucked into a discussion about test > methodologies, so I'm happy to concede the point if it will get me out > of that debate. > > Alright, I'll see if I can find my script somewhere and copy+paste it > to make a bugreport, but it's a pretty obnoxious bug report. It's a > write-once throwaway script that probably has all sorts of stuff wrong > with it. Also, it takes as input `git log` from the linux kernel git > repo, which is pretty large. > > Once, while running it, an exception escaped from a try/catch and > killed the script. But it only happened once so I don't know how many > times you'd have to re-run it to get that result. So, that's not > really nicely reproducible. > > Otherwise, if you remove the try/catch statements a couple of string > related things will blow up with an exception. > > The entitled brat response wasn't aimed at you (Jeff), but I've > literally never written anything negative about an open source project > without having someone tell me that I'm an entitled jerk, so I > expected to get that response to this post. And I did, so that streak > continues! > > On Mon, Dec 29, 2014 at 4:37 PM, Patrick O'Leary > <[email protected]> wrote: > > On Monday, December 29, 2014 4:13:35 PM UTC-6, Jeff Bezanson wrote: > >> > >> But one > >> feature we could stand to add is asserting properties that must be > >> true for all arguments, and running through lots of combinations of > >> instances. > > > > > > Anyone who is interested in this is welcome to use > > https://github.com/pao/QuickCheck.jl as a starting point. >
