Capture failed pipeline

[pmatos]

I am starting to have more and more scripts using rash but they are quite flaky in general. One thing I still haven't figured out the best way to achieve is grab a pipeline failure with the failing exit code. For example:

#lang rash
(define (my-false)
  {true
   echo "1"
   false
   echo "42"})

I get:

> (my-false)
1
; run-pipeline: unix pipeline-segment (#<path:/usr/bin/false>) terminated with code 1
; 
; Context:
;  /home/pmatos/.racket/7.3/pkgs/shell-pipeline/private/mixed-pipeline.rkt:161:6 for-loop
;  /home/pmatos/.racket/7.3/pkgs/shell-pipeline/private/mixed-pipeline.rkt:257:0 unpack21
;  /home/pmatos/Projects/igalia/jsc-utils/webkit.rkt:23:0 my-false
;  /home/pmatos/installs/racket-7.3/collects/racket/repl.rkt:11:26

I see the error call that terminates this in the rash source (https://github.com/willghatch/racket-rash/blob/master/shell-pipeline/private/mixed-pipeline.rkt) however this error call is less than ideal if I want to grab the error and the exit code. I could in principle catch the exception thrown by error (exn:fail) and parse the string, however this is less than ideal. Is there a good way to actually do this at the moment?

[willghatch]

On Mon, Jun 24, 2019 at 06:16:49AM -0700, Paulo Matos wrote:

I am starting to have more and more scripts using rash but they are quite flaky in general.

Cool! Except for them being flaky...

One thing I still haven't figured out the best way to achieve is grab a pipeline failure with the failing exit code. For example:

#lang rash
(define (my-false)
 {true
  echo "1"
  false
  echo "42"})

I get:

> (my-false)
1
; run-pipeline: unix pipeline-segment (#<path:/usr/bin/false>) terminated with code 1
;
; Context:
;  /home/pmatos/.racket/7.3/pkgs/shell-pipeline/private/mixed-pipeline.rkt:161:6 for-loop
;  /home/pmatos/.racket/7.3/pkgs/shell-pipeline/private/mixed-pipeline.rkt:257:0 unpack21
;  /home/pmatos/Projects/igalia/jsc-utils/webkit.rkt:23:0 my-false
;  /home/pmatos/installs/racket-7.3/collects/racket/repl.rkt:11:26

I see the error call that terminates this in the rash source (https://github.com/willghatch/racket-rash/blob/master/shell-pipeline/private/mixed-pipeline.rkt) however this error call is less than ideal if I want to grab the error and the exit code. I could in principle catch the exception thrown by error (exn:fail) and parse the string, however this is less than ideal. Is there a good way to actually do this at the moment?

Yeah, that's one of the many things that while implementing I decided I would improve later. In this case I think it just needs a new exception type, something like exn:unix-pipeline-member-failure, that stores the return value and stderr text. Or is there anything else that you think would be better?

Feel free to send me a pull request. Or I'm also happy to just implement it.

What else is causing flakiness, or how is flakiness manifested in your scripts?

[pmatos]

The flakiness comes exactly from this problem of not being able to properly catch failures. A new exception is exactly what I was thinking about. I will take a look at it later today and if I can put together something I will send a PR even if in draft mode. Glad you agree this can be improved. Thanks for rash - it's awesome - needs a marketing team!