Changes from v1:
1. (Thanks flokli for noticing!): first patch (pcsclite: remove isLinux
condition from --enable-ipcdir) of previous version would produce undesired
results (enabled `--with-systemunitdir` everywhere) if applied without the
second patch in the series. Fixed in v2. No change to the result of
applying the entire series.
2. `pcsclite: make udevSupport optional`: previous series swapped the order of
two of the buildInputs, which would cause an unnecessary rebuild in the
common (hydra) case. This version maintains the original order, so no
rebuild.
[PATCH 1/6] pcsclite: remove isLinux condition from --enable-ipcdir configureFlag
I tried building `pkgsCross.aarch64-multiplatform.pcsc-tools` from
a `x86_64-linux` system.
At least on master, there was a cross regression for gnupg (which I hit
when building pcsc-tools with systemd support, as systemd (or importd more
specifically) uses gnupg.
Though if I disable that, it successfully builds, it's unrelated to
your patchset anyways (and maybe fixed in staging, didn't check).
On Mon, Apr 15, 2024 at 01:09:16PM -0700, Adam Joseph wrote:
>Changes from v1:>>1. (Thanks flokli for noticing!): first patch (pcsclite: remove isLinux> condition from --enable-ipcdir) of previous version would produce undesired> results (enabled `--with-systemunitdir` everywhere) if applied without the> second patch in the series. Fixed in v2. No change to the result of> applying the entire series.>>2. `pcsclite: make udevSupport optional`: previous series swapped the order of> two of the buildInputs, which would cause an unnecessary rebuild in the> common (hydra) case. This version maintains the original order, so no> rebuild.
Note this still causes a lot of rebuilds, probably due to the move of
gobject-introspection (?).
Are you OK with this simply targeting staging?
flokli
On Tue, Apr 16, 2024 at 10:57:24PM +0300, Florian Klink wrote:
>I tried building `pkgsCross.aarch64-multiplatform.pcsc-tools` from>a `x86_64-linux` system.>>At least on master, there was a cross regression for gnupg (which I hit>when building pcsc-tools with systemd support, as systemd (or importd more>specifically) uses gnupg.>>Though if I disable that, it successfully builds, it's unrelated to>your patchset anyways (and maybe fixed in staging, didn't check).>>On Mon, Apr 15, 2024 at 01:09:16PM -0700, Adam Joseph wrote:>>Changes from v1:>>>>1. (Thanks flokli for noticing!): first patch (pcsclite: remove isLinux>> condition from --enable-ipcdir) of previous version would produce undesired>> results (enabled `--with-systemunitdir` everywhere) if applied without the>> second patch in the series. Fixed in v2. No change to the result of>> applying the entire series.>>>>2. `pcsclite: make udevSupport optional`: previous series swapped the order of>> two of the buildInputs, which would cause an unnecessary rebuild in the>> common (hydra) case. This version maintains the original order, so no>> rebuild.>>Note this still causes a lot of rebuilds, probably due to the move of>gobject-introspection (?).>Are you OK with this simply targeting staging?
As the patches started to not apply anymore, I manually rebased this
on staging, and sent it out (modulo some formatter fixes) to
https://github.com/NixOS/nixpkgs/pull/305260.
I'll undraft this PR once I finished building.
Quoting Florian Klink (2024-04-19 01:23:23)
> >Note this still causes a lot of rebuilds, probably due to the move of> >gobject-introspection (?).> >Are you OK with this simply targeting staging?
Yes of course; sorry about the slow reply.
> As the patches started to not apply anymore, I manually rebased this> on staging, and sent it out (modulo some formatter fixes) to> https://github.com/NixOS/nixpkgs/pull/305260.>> I'll undraft this PR once I finished building.
I see that you did, and that it was merged. Thank you so much!
- a