#general
- r
rough-minister-58256
09/09/2015, 6:31 AMcuz why else would you just run ‘pex’ then - r
rough-minister-58256
09/09/2015, 6:31 AMunless you thought you had just produced a pex - e
enough-analyst-54434
09/09/2015, 6:31 AMpex something pex -> a pex that when run and drops interpreter and doesn't give you something will confuse me every time - e
enough-analyst-54434
09/09/2015, 6:31 AMEven though I understand it now - r
rough-minister-58256
09/09/2015, 6:33 AMi have many pexs that do exactly that, one with a collection of various data science libs etc w/ pandas/numpy/scipy/matplotlib et al - r
rough-minister-58256
09/09/2015, 6:33 AMwould be odd to invoke that pex with —repl - r
rough-minister-58256
09/09/2015, 6:33 AMetc - r
rough-minister-58256
09/09/2015, 6:33 AMthe default behavior of python is to drop you into a repl, so python-with-stuff should probably have that same default behavior - r
rough-minister-58256
09/09/2015, 6:34 AMi think for collections of deps it makes total sense - e
enough-analyst-54434
09/09/2015, 6:34 AMYes, it's only only when stuff includes pex itself and you said -c pex that is confusing - e
enough-analyst-54434
09/09/2015, 6:34 AMOnly that - r
rough-minister-58256
09/09/2015, 6:34 AMwhat entrypoint do you give somethign that contains a toolkit etc - r
rough-minister-58256
09/09/2015, 6:35 AMya, perhaps the better thing here is to discourage ppl from bootstrapping their own pex as the first thing in the doc - e
enough-analyst-54434
09/09/2015, 6:35 AMI think it's that and or simpler make no args yell! - r
rough-minister-58256
09/09/2015, 6:36 AMor both - e
enough-analyst-54434
09/09/2015, 6:36 AMThe issue is not bootstrapping pex, it's adding in another thing too - e
enough-analyst-54434
09/09/2015, 6:36 AMThat's what confused Dan and I - e
enough-analyst-54434
09/09/2015, 6:37 AMYou make prebuilt env pex interpreters all the time and his example looks like that, but it's not - e
enough-analyst-54434
09/09/2015, 6:37 AMBecause of the -c with funky no args behavior - e
enough-analyst-54434
09/09/2015, 6:37 AMThat acts just like a pex with no enyrypoint... almost - r
rough-minister-58256
09/09/2015, 6:38 AMits really just that it mimics the standard entrypoint - and for good reason, its invoking itself! - e
enough-analyst-54434
09/09/2015, 6:39 AMHere's my issue - I think you've helped me understand it - e
enough-analyst-54434
09/09/2015, 6:43 AMIt's all down to the -c, it's easy to forget that the pex environment with adjoined fabric say, has an entry point after you've built it. You list the zip - there's requests, You run the pex and it drops you to an interp - which for all other pex es means you never defined an entrypoint - e
enough-analyst-54434
09/09/2015, 6:44 AMSo any pex that includes pex with -c , run 1 year later looks like a pex environment with no entrypoints - r
rough-minister-58256
09/09/2015, 6:44 AMunless you provide it some args - r
rough-minister-58256
09/09/2015, 6:44 AMor -h - r
rough-minister-58256
09/09/2015, 6:44 AMbut yeah - e
enough-analyst-54434
09/09/2015, 6:45 AMBut the dissonance comes when that outward appearance blows up and impotent from the environment fails - e
enough-analyst-54434
09/09/2015, 6:46 AMI think your solution of just changing the example is probably the way to go - e
enough-analyst-54434
09/09/2015, 6:46 AMI posit without that example, no one who would be confused by it, like Dan and me, would ever try to build such a pex in the 1st place