Upgrade to tar v3 #1212
Upgrade to tar v3 #1212
Conversation
|
Note: This drops support for node 0.x, and thus should be considered a breaking semver-major change, and not landed until those node versions are no longer supported. |
|
@bnoordhuis I see at the LTS schedule that 0.12 was only in "maintenance" through 2016-12-31. I think that means it isn't nominally supported? Still a major version bump, of course. |
|
0.12 is no longer supported by Node.js, whether or not 0.12 is still supported by the current version of node-gyp, I am not really sure. |
|
I’d say you can’t just remove 0.x support in a non-major release; too many CI configurations that use |
|
I suppose that now that node 8 is out, we might as well bump major and drop support for node < 4. There still seem to be quite a few v0.10.x users but those people probably don't upgrade frequently in general and node-gyp in particular. |
|
Is there an LTS/maintenance plan for BTW: to the package owners we need a |
|
@isaacs IMHO this PR should come with a |
|
Ehhhh, if people really want a 3.x maintenance we'll probably only know if they complain about it so I'd say leave it be and consider doing something if people post issues about that specifically. How do we manage these around here? Merge everything into |
Agreed
Might as well stay on master, we could create a |
I don't know of a |
|
You all could probably check the nodejs.org log files to see how many 0.x users are downloading gyp headers, but here at npm, we see that 0.10 and 0.12 together account for a little less than 6% of npm registry traffic, and they're almost all using older versions of npm (and thus, won't get a major version bump of node-gyp). All other 0.x node versions make up less than 0.1% of npm registry traffic. In other words, it's almost certainly safe to drop support for 0.x in a major version bump. |
|
@isaacs if I read this thread correctly there's consensus on dropping < 4 support in a major version bump. AFAICT We just don't have an established procedure on how to maintain two version lines of a package (only personal experience). IMHO it does not need to be as strict as I say we just wing it 😉 since collectively we have plenty of prior experience in maintaining packages. |
Tar version 3 performs better and is more well tested than its predecessor. npm will be using this in the near future, so there is no benefit in shipping a node-gyp that uses the slower and less reliable fstream-based tar. This drops support for node 0.x, and thus should be considered a breaking semver-major change.
|
@refack updated package.json engines field to specify >= 4 |
10x, lets see how much the CI is broken |
Tar version 3 performs better and is more well tested than its predecessor. npm will be using this in the near future, so there is no benefit in shipping a node-gyp that uses the slower and less reliable fstream-based tar. This drops support for node 0.x, and thus should be considered a breaking semver-major change. PR-URL: nodejs#1212 Reviewed-By: Refael Ackermann <refack@gmail.com> Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl> Reviewed-By: Gibson Fahnestock <gibfahn@gmail.com>
Tar version 3 performs better and is more well tested than its predecessor. npm will be using this in the near future, so there is no benefit in shipping a node-gyp that uses the slower and less reliable fstream-based tar. This drops support for node 0.x, and thus should be considered a breaking semver-major change. PR-URL: nodejs#1212 Reviewed-By: Refael Ackermann <refack@gmail.com> Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl> Reviewed-By: Gibson Fahnestock <gibfahn@gmail.com>
Tar version 3 performs better and is more well tested than its predecessor. npm will be using this in the near future, so there is no benefit in shipping a node-gyp that uses the slower and less reliable fstream-based tar. This drops support for node 0.x, and thus should be considered a breaking semver-major change. PR-URL: #1212 Reviewed-By: Refael Ackermann <refack@gmail.com> Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl> Reviewed-By: Gibson Fahnestock <gibfahn@gmail.com>
Tar version 3 performs better and is more well tested than its
predecessor. npm will be using this in the near future, so there is no
benefit in shipping a node-gyp that uses the slower and less reliable
fstream-based tar.