Ask HN: Do we have a package manager problem?
Nowadays, every big software project seems to need its own package manager: programming languages, os distributions, virtualization and sandboxing solutions, text editors, programs with plugins... Do they really all have significant different needs? Will we ever have the 'git' of package managers?
2 comments
It's not generally that bad unless you want the latest and greatest. For example, on Debian, you can easily install Perl modules from APT instead of CPAN (which I've found temperamental in practise). The only trouble is that the modules are often older than those you can get through the dedicated PM channel (since obviously the maintainers have to package up their release in line with Debian standards).
Package management is still a much better idea than manual dependency installation - even NuGet for .Net is an improvement as it supports automatic restoration of dependencies at build time. However, as we are immersed in so many different platforms and technologies, it is extremely difficult to have a one-size-fits-all PM. Linux PMs are probably the closest.
Package management is still a much better idea than manual dependency installation - even NuGet for .Net is an improvement as it supports automatic restoration of dependencies at build time. However, as we are immersed in so many different platforms and technologies, it is extremely difficult to have a one-size-fits-all PM. Linux PMs are probably the closest.
What is "the git of package managers"? One thing that a very loud group of people claim is the second coming of jesus without any knowledge of the alternatives?
I think we have that already with NPM.
Edit: to clarify, I'm not suggesting npm is the 2nd coming. I'm suggesting people act like it is, ( much as they do with git )
I think we have that already with NPM.
Edit: to clarify, I'm not suggesting npm is the 2nd coming. I'm suggesting people act like it is, ( much as they do with git )
Then i could say we have that already with Rubygems.
I always cry a little when a nice looking small tool is distributed trough NPM. Why? I really dont want and will build those dependencies just for a stupid mini tool.
I always cry a little when a nice looking small tool is distributed trough NPM. Why? I really dont want and will build those dependencies just for a stupid mini tool.
To me 'the git of *' is something:
- based on good theoretical foundations
- sufficiently performant
- not perfect, but good enough for most use cases
- accepted as a de-facto standard, but that is a consequence of the above