Hacker News new | comments | show | ask | jobs | submit login

It's clear from these comments that I need to explain the project more concisely on the blog (which I mentioned in the conclusion), but here is a short response:

(1) In retrospect, I should have responded to the first commenter differently. He was clearly angered by having to debug other people's shell scripts in the past. But that is exactly what I'm trying to fix -- I'm trying to provide an UPGRADE PATH OUT OF BASH. I explicitly state that in this long post, which I think some people have missed.

These posts show one part of the plan: http://www.oilshell.org/blog/tags.html?tag=osh-to-oil#osh-to...

(2) I'll repeat my objection: trying to convince people not to use bash is going to be about as successful as convincing people not to use PHP. Not only is it not a solution to the problem, it also ignores the humongous installed base of PHP code, like Wikipedia, etc.

Even if another line of bash never gets written, you'll still have to deal with it on a regular basis!

I am providing a path out of bash, while other people are just wishing the problem would go away. The world is not how you want it to be, but exhortation on online forums isn't going to change the world. Less charitably, "you're wasting your breath".

I need to write a blog post entitled "Reimplementing Bash is the Only Way to Kill It". This is analogous to how Facebook is "removing" PHP from their codebase by developing the Hack language and VM.

It's also analogous to how Microsoft "killed" Lotus and other competitors by understanding their file formats.

(3) I also don't think you understand why cloud projects like Kubernetes are using shell scripts, probably because you don't work in that domain. Shell really is the best tool for that job.

I think it's presumptuous to state that the people writing that code don't know what they're doing. A lot of the project is in Go, but a lot of it is in shell, and there's probably a reason for that other than ignorance.




To start, I think your project is excellent and I don't mean to take away from your work. It has bore fruit, which is an accomplishment on it's own, to say nothing of the intrinsic value of simply investigating and hacking in whatever domain interests you. I like to think I'm fond of hacking in "unsexy" spaces, and I think hot-rodding a shell is that space.

Re: 3) above -- Curious to hear about the reasons.

Re: "...people writing that code ...other than ignorance" -- I'm not trying to imply people working on this are idiots... though w/ broad-sounding statements like "scripts with more than 20 lines should be looked at suspiciously" I can see how someone might read that. I think there are myriad poor reasons an shell script might be inappropriately used. There's also a case for "if it's not mine, who am I to judge?". I'd be curious to hear points of why shell is especially suited for (e.g.) cloud projects like Kubernetes, though.

Good luck with your work.


For #3, shell is a language for dealing with processes and the file system. And that's exactly what you need to bring up servers / containers / cluster managers like Kubernetes. Also stuff like OpenStack and distributed file systems. Shell is universally used to automate these projects.

I think you need to actually do it to have an appreciation for this. Other people here have made the same points I'm making, but if you don't try it, you won't viscerally understand it. I would just caution against the attitude of assuming that people who use a tool you don't like / don't want to learn (understandably) are using the wrong tool for the job.

As mentioned I should write a blog post called "Python Is Not an Acceptable Shell", in part based on this post, which is very in detail, but I think proves the opposite point:

https://medium.com/capital-one-developers/bashing-the-bash-r...

Of course, bash has significant downsides for this task too. That is why Oil exists!




Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact

Search: