Fix Archlinux package creation #425

2013-08-21T00:12:24+02:00

rakoo commented

2013-08-21 00:12:24 +02:00

(Migrated from github.com)

When simply cloning from the repo, the folder name is PyBitmessage, not just pybitmessage.
Building the xxx-pkg.tar.xz is the job of Archlinux's makepkg, not the job of upstream (ie bitmessage). You just need to build the xxx.tar.gz part

An archer still has to manually clone the repo to install it with the provided PKGBUILD. Is there any reason why you don't provide xxx.tar.gz-ified extracts of the code on each release ? (I also notice that you don't use tags, which could be really helpful with github). This would greatly help, because then all you'd need to distribute is the PKGBUILD along with an url pointing to the xxx.tar.gz (I think github hosts them automatically)

- When simply cloning from the repo, the folder name is PyBitmessage, not just pybitmessage. - Building the `xxx-pkg.tar.xz` is the job of Archlinux's `makepkg`, not the job of upstream (ie bitmessage). You just need to build the `xxx.tar.gz` part An archer still has to manually clone the repo to install it with the provided PKGBUILD. Is there any reason why you don't provide `xxx.tar.gz`-ified extracts of the code on each release ? (I also notice that you don't use tags, which could be really helpful with github). This would greatly help, because then all you'd need to distribute is the PKGBUILD along with an url pointing to the `xxx.tar.gz` (I think github hosts them automatically)

Atheros1 commented

2013-08-23 23:16:47 +02:00

(Migrated from github.com)

I am tagging releases from now on.

Atheros1 commented

2013-08-23 23:27:50 +02:00

(Migrated from github.com)

An archer still has to manually clone the repo

I think that it is assumed that you already have the source code either from using 'git clone' or from using the 'download zip' button on Github. You'll get the latest code this way. But I'm not very familiar with Arch packages.

> An archer still has to manually clone the repo I think that it is assumed that you already have the source code either from using 'git clone' or from using the 'download zip' button on Github. You'll get the latest code this way. But I'm not very familiar with Arch packages.

fiatflux commented

2013-08-23 23:58:31 +02:00

(Migrated from github.com)

I've already complained about, and conceded to, the arch stuff here. But I think it's telling that arch users didn't make it.

Regarding tags, latest code is not always desired. C.f. those fancy >>>>>> and <<<<<< things from unresolved merge conflicts :-)

I've already complained about, and conceded to, the arch stuff here. But I think it's telling that arch users didn't make it. Regarding tags, latest code is not always desired. C.f. those fancy >>>>>> and <<<<<< things from unresolved merge conflicts :-)

fiatflux commented

2013-08-24 00:28:29 +02:00

(Migrated from github.com)

I've already complained about, and conceded to, the arch stuff here. But I
think it's telling that arch users didn't make it.

Regarding tags, latest code is not always desired. C.f. those fancy >>>>>>
and <<<<<< things from unresolved merge conflicts :-)

On Fri, Aug 23, 2013 at 11:27 PM, Jonathan Warren
notifications@github.comwrote:

An archer still has to manually clone the repo

I think that it is assumed that you already have the source code either
from using 'git clone' or from using the 'download zip' button on Github.
You'll get the latest code this way.

—
Reply to this email directly or view it on GitHubhttps://github.com/Bitmessage/PyBitmessage/pull/425#issuecomment-23192220
.

I've already complained about, and conceded to, the arch stuff here. But I think it's telling that arch users didn't make it. Regarding tags, latest code is not always desired. C.f. those fancy >>>>>> and <<<<<< things from unresolved merge conflicts :-) On Fri, Aug 23, 2013 at 11:27 PM, Jonathan Warren notifications@github.comwrote: > An archer still has to manually clone the repo > > I think that it is assumed that you already have the source code either > from using 'git clone' or from using the 'download zip' button on Github. > You'll get the latest code this way. > > — > Reply to this email directly or view it on GitHubhttps://github.com/Bitmessage/PyBitmessage/pull/425#issuecomment-23192220 > .

Atheros1 commented

2013-08-25 22:16:14 +02:00

(Migrated from github.com)

Thank you rakoo and fiatflux.

rakoo commented

2013-08-26 01:04:06 +02:00

(Migrated from github.com)

But I'm not very familiar with Arch packages.

Here's a very rough explanation:

To create a package, Arch basically needs a PKGBUILD. This file is more or less an improved Makefile. In the PKGBUILD, one of the important variable is sources, which tells the packager where the source code can be retrieved. It usually is an http url, or sometimes a git url, so that it can be retrieved on the internet. The packager is not supposed to manually download the code.

Regarding PyBitmessage, a PKGBUILD could contain the link to the latest code, ie the link to the download button... but I'd like to use more stable releases (and I see you already have releases, so to provide releases url it should be a matter of simply tagging them in git, and github automatically creates a download button for it)

> But I'm not very familiar with Arch packages. Here's a very rough explanation: To create a package, Arch basically needs a PKGBUILD. This file is more or less an improved Makefile. In the PKGBUILD, one of the important variable is `sources`, which tells the packager where the source code can be retrieved. It usually is an `http` url, or sometimes a `git` url, so that it can be retrieved on the internet. The packager is not supposed to manually download the code. Regarding PyBitmessage, a PKGBUILD _could_ contain the link to the latest code, ie the link to the download button... but I'd like to use more stable releases (and I see you already have releases, so to provide releases url it should be a matter of simply tagging them in git, and github automatically creates a download button for it)

This repo is archived. You cannot comment on pull requests.

No reviewers