search

AOSC-Archive / autobuild3

AOSC OS package maintenance toolkit (version 3)
https://aosc.io
GNU General Public License v2.0
24 stars 17 forks source link

Please drop RPMEPOCH support #79

Closed Artoria2e5 closed 8 years ago

Artoria2e5 commented 8 years ago

The current RPMEPOCH design is a leftover before ab2-port introduction.

When I designed the PKGEPOCH in ab2-port, it was intended to be a fully rpm-compliant epoch, so it was only a bash integer, without anything fancy like makepkg floats. Yes, no need for a separate epoch.

Additionally, using a separate EPOCH lead to inconsistent outputs from pm_getver, since RPM only knows about the epoch being written in the spec.

Solution: In pm.sh, declare -n RPMEPOCH=PKGEPOCH.

Icenowy commented 8 years ago

The tradition epoch that I said is: If you do not attach such an epoch, your package will be denied by rpm, and your system won't be LSB-complicant

---- Mingye Wang编写 ----

The current RPMEPOCH design is a leftover before ab2-port introduction.

When I designed the PKGEPOCH in ab2-port, it was intended to be a fully rpm-compliant epoch, so it was only a bash integer, without anything fancy like makepkg floats.

Additionally, using a separate EPOCH lead to inconsistent outputs from pm_getver, since RPM only knows about the epoch being written in the spec.

Solution: In pm.sh, declare -n RPMEPOCH=PKGEPOCH.

Reply to this email directly or view it on GitHub: https://github.com/AOSC-Dev/autobuild3/issues/79

MingcongBai commented 8 years ago

Yes, and perl is the classic case in the discussion. 2015年12月17日 下午9:42，"Icenowy Zheng" notifications@github.com写道：

The tradition epoch that I said is: If you do not attach such an epoch, your package will be denied by rpm, and your system won't be LSB-complicant

  • 发送自我的Sony Xperia™智能手机

---- Mingye Wang编写 ----

The current RPMEPOCH design is a leftover before ab2-port introduction.

When I designed the PKGEPOCH in ab2-port, it was intended to be a fully rpm-compliant epoch, so it was only a bash integer, without anything fancy like makepkg floats.

Additionally, using a separate EPOCH lead to inconsistent outputs from pm_getver, since RPM only knows about the epoch being written in the spec.

Solution: In pm.sh, declare -n RPMEPOCH=PKGEPOCH.

Reply to this email directly or view it on GitHub: https://github.com/AOSC-Dev/autobuild3/issues/79

— Reply to this email directly or view it on GitHub https://github.com/AOSC-Dev/autobuild3/issues/79#issuecomment-165672221.

Artoria2e5 commented 8 years ago

So just make a declare -i PKGEPOCH in pmlib. This is fscking easy.

MingcongBai commented 8 years ago

The problem is do we want the hacky bit that RPM requires to be found on DPKG as well... 2015年12月17日 下午9:44，"Mingye Wang" notifications@github.com写道：

So just make a declare -i PKGEPOCH in pmlib.

— Reply to this email directly or view it on GitHub https://github.com/AOSC-Dev/autobuild3/issues/79#issuecomment-165672378.

Artoria2e5 commented 8 years ago

The problem is with a fscking inconsistent output from pm_getver, your dependency info generated is fscking bootstrap.

Icenowy commented 8 years ago

Here's a solution: add the global epoch to rpm levels

---- Mingye Wang编写 ----

The problem is with a fscking inconsistent output from pm_getver, your dependency info generated is fscking bootstrap.

Reply to this email directly or view it on GitHub: https://github.com/AOSC-Dev/autobuild3/issues/79#issuecomment-165672516

MingcongBai commented 8 years ago

Very true. But calm down. A unity on version payload is not a issue on functionality anyways, I mean, I can live with an extra epoch can't I... 2015年12月17日 下午9:46，"Mingye Wang" notifications@github.com写道：

The problem is with a fscking inconsistent output from pm_getver, your dependency info generated is fscking bootstrap.

— Reply to this email directly or view it on GitHub https://github.com/AOSC-Dev/autobuild3/issues/79#issuecomment-165672516.

Artoria2e5 commented 8 years ago

Just make them the same. Already said that in comment 0.

Artoria2e5 commented 8 years ago

Design failure is design failure. Remove it before people get used to it.

MingcongBai commented 8 years ago

Explain that a little further? 2015年12月17日 下午9:48，"Icenowy Zheng" notifications@github.com写道：

Here's a solution: add the global epoch to rpm levels

  • 发送自我的Sony Xperia™智能手机

---- Mingye Wang编写 ----

The problem is with a fscking inconsistent output from pm_getver, your dependency info generated is fscking bootstrap.

Reply to this email directly or view it on GitHub: https://github.com/AOSC-Dev/autobuild3/issues/79#issuecomment-165672516

— Reply to this email directly or view it on GitHub https://github.com/AOSC-Dev/autobuild3/issues/79#issuecomment-165672731.

MingcongBai commented 8 years ago

I will agree with you on this point. Please proceed. 2015年12月17日 下午9:50，"Mingye Wang" notifications@github.com写道：

Design failure is design failure. Remove it before people get used to it.

— Reply to this email directly or view it on GitHub https://github.com/AOSC-Dev/autobuild3/issues/79#issuecomment-165672939.

Artoria2e5 commented 8 years ago

Too lazy to touch the code. All solutions are there, just apply them yourself.

I am wiping the OpenJDK butt.

MingcongBai commented 8 years ago

Alright, tomorrow morning. 2015年12月17日 下午9:57，"Mingye Wang" notifications@github.com写道：

Too lazy to touch the code. All solutions are there, just apply them yourself.

I am wiping the OpenJDK butt.

— Reply to this email directly or view it on GitHub https://github.com/AOSC-Dev/autobuild3/issues/79#issuecomment-165673562.

Artoria2e5 commented 8 years ago

?