search

npm / registry-issue-archive

An archive of the old npm registry issue tracker
https://npm.community
249 stars 47 forks source link

users would like to be able to republish the readme without bumping their package version #46

Open snopeks opened 8 years ago

snopeks commented 8 years ago

Based on user feedback, we're hearing it would be nice for users to republish/refresh their readme without bumping their package version.

See example feedback: https://twitter.com/mrvautin/status/767184966564274176

screen shot 2016-08-22 at 5 35 19 pm

Thoughts?

kemitchell commented 8 years ago

Three thoughts:

  1. There's not much cost to doc-only patch releases. ^ ranges are default, and some more conservative folk use ~. Either picks up patch releases, including doc-only patch releases. npm i is unlikely to duplicate a package with versions only a single patch version apart.
  2. I'm in the (admittedly idiosyncratic) habit of putting my tests in README. A script extracts the fenced code blocks and pipes to Node.js.
  3. I take it a hypothetical npm update-readme wouldn't affect the README clients receive in tarball on npm install. I fully expect most folks peruse READMEs on npmjs.com and github.com, rather than less like this old fogey. But mismatches could confuse, and potentially confound if, say, the README online has license information and the README in tarball does not.
mrvautin commented 8 years ago

Some really valid points there. I ignorantly assumed that because I consume the README via the registry that everyone else does this too.

Seems like you killed that idea quick smart! 😃

kemitchell commented 8 years ago

@mrvautin ideas can't be killed, and I wouldn't try if they could! God help anyone who gives me the last word on matters npm-ish.

As for READMEs on npmjs.com, again, I think I'm in a small minority that looks in node_modules first. If it were a simple matter of pleasing either browser-folk or pager-folk, I'd expect npm CLI and npm, Inc. to favor browser-folk.

mrvautin commented 8 years ago

@kemitchell I hope you didn't take offence. I was referring to your points being a very good reason to leave it the way it is. The risk of doco mismatch probably outweighs the convenience.

wesleytodd commented 7 years ago

I think a fairly strong case could be made for a update-readme command which would follow @kemitchell's reccomendation of not updating the tarball, but updates the website display. I often release x.x.1 versions just to update formatting or typo's in the readme.