Re: Eternal Terminal Server?
Hey yasu, I've never actually heard of eternal terminal, but that sounds pretty awesome! haha! If those guix packages are only for clients, then it sounds like we need to create a package for the server stuff, and also create a service for it. Would you care to work on that? I can probably give you encouragement going forward, but probably not much developmental guidance. :) -- Joshua Branson Sent from Emacs and Gnus https://gnucode.me https://video.hardlimit.com/accounts/joshua_branson/video-channels "You can have whatever you want, as long as you help enough other people get what they want." - Zig Ziglar
Re: Touchpad not working on Acer Aspire One laptop
Hey Barnabas! (awesome name by the way!) Is your touchpad working at all? Can you move a mouse with your touchpad? I had an issue on my old macbook, where my mouse would only move up and down. When I started using sway, that issue went away. Thanks, -- Joshua Branson Sent from Emacs and Gnus https://gnucode.me https://video.hardlimit.com/accounts/joshua_branson/video-channels "You can have whatever you want, as long as you help enough other people get what they want." - Zig Ziglar
Re: Trouble booting Guix System
Yeah, any AMD graphics card is going to have a hard time. Without the firmware files, those graphics cards are nearly useless. There's some really old nvidia cards that work fairly well. There's probably a kernel parameter you could pass that would disable the AMD graphics driver from loading. The Intel CPU graphics driver is one of the better ones for running libre graphics software. :) -- Joshua Branson Sent from Emacs and Gnus https://gnucode.me https://video.hardlimit.com/accounts/joshua_branson/video-channels "You can have whatever you want, as long as you help enough other people get what they want." - Zig Ziglar
Re: Touchpad not working on Acer Aspire One laptop
On Tue, Nov 10, 2020 at 03:05:05PM +0100, Barnabás Béres wrote: > Hi all, > > I successfully installed Guix on a Acer Aspire One laptop but the touchpad > not working. Do you have any idea how I can make it work. This model has a > SynPS/2 Synaptics TouchPad. Can you share your config.scm? That will help us give advice.
Confusing typo (was: Easy DigitalOcean setup?)
David Dashyan writes: > I've recently mutated one of Digitalocean debian machines using method > provided by Reza in this thread I think there . I would check > bootloader config -- I think you should have "/dev/vda" there. Missed this typo, sorry… It should have been: I've recently mutated one of Digitalocean debian machines using method provided by Reza in this thread. I would check bootloader config -- I think you should have "/dev/vda" there. -- David aka zzappie
Re: Easy DigitalOcean setup?
Hello raingloom! I've recently mutated one of Digitalocean debian machines using method provided by Reza in this thread I think there . I would check bootloader config -- I think you should have "/dev/vda" there. By the way Jérémy! Did It work for you with dhcp-client??? I thought you need to use static networking. I am asking because I've recently was automating droplets deployment by quering digitalocean's metadata service to find out the network configuration. P.S. yggdrasil! That is an interesting piece of software! -- David aka zzappie signature.asc Description: PGP signature
Re: How do I correctly relocate PostGIS control files?
Carlo Zancanaro writes:
> I didn't respond initially because I don't have any specialist
> knowledge about this. I had some free time today, so I did a bit
> of an investigation and I think I've figured out what's gone wrong for
> you here.
>
> In your original email you gave this definition for postgresql-13:
>
> ...
>
> However, this is missing one important line from the original
> postgresql definition, which applies a patch to the source:
>
>(patches (search-patches
>"postgresql-disable-resolve_symlinks.patch"))
>
> ...
>
> I hope that helps!
***
All hail the mighty Carlo Zancanaro!
Applying that patch to my postgresql-13 package does indeed fix the issue!
***
I have also been digging through the code behind
postgresql-service-type, and I worked out what I'm guessing is the
reason that patch actually works.
In gnu/services/databases.scm, postgresql-service-type is declared and
uses the following function to compute the postgresql package to use for
both its database activation (postgresql-activation) and its shepherd
service (postgresql-shepherd-service):
(define (final-postgresql postgresql extension-packages)
(if (null? extension-packages)
postgresql
(package
(inherit postgresql)
(source #f)
(build-system trivial-build-system)
(arguments
`(#:modules ((guix build utils) (guix build union))
#:builder
(begin
(use-modules (guix build utils) (guix build union) (srfi srfi-26))
(union-build (assoc-ref %outputs "out") (map (lambda (input) (cdr
input)) %build-inputs))
#t)))
(inputs
`(("postgresql" ,postgresql)
,@(map (lambda (extension) (list "extension" extension))
extension-packages))
When extension-packages is (list postgis), then final-postgresql creates
a union-build of both the postgresql and postgis packages, which
essentially just creates a new /gnu/store directory containing the
outputs of both packages.
In exploring my /gnu/store directory, I noticed that even before I
applied Carlo's patch code, there already existed a union-build of my
postgresql-13 and postgis-for-postgresql-13 packages. However, checking
my process table with ps showed that the postgres executable started by
the shepherd service was coming from a /gnu/store directory that ONLY
contained postgresql-13 and NOT from the union-build directory as I
would have expected from the final-postgresql function definition above.
So what gives?
It looks like the individual postgresql-13 and postgis-for-postgresql-13
packages were built correctly and the union-build was also created
correctly (probably either at activation time or at service start time).
HOWEVER, the union-build directory consists primarily of symlinks to the
files in the individual postgresql-13 and postgis-for-postgresql-13
store directories. My guess then is that when the shepherd service fired
up the postgres executable, it was trying to run it from the union-build
directory but after following the symlink, it reported in ps the path to
the target of that symlink rather than the location of the symlink
itself. Then postgres must have taken this symlink target directory as
the actual directory is was being run from, which of course did not
contain any postgis control files.
Long story short, Carlo noticed that the default postgresql package in
gnu/packages/databases.scm, applied this patch to its source code:
postgresql-disable-resolve_symlinks.patch:
***
>From 223c82d1d6ed1f29f26307249827ff679e09c780 Mon Sep 17 00:00:00 2001
From: Julien Lepiller
Subject: [PATCH] disable resolve_symlink
---
src/common/exec.c | 2 ++
1 file changed, 2 insertions(+)
diff --git a/src/common/exec.c b/src/common/exec.c
index 878fc29..6b3e283 100644
--- a/src/common/exec.c
+++ b/src/common/exec.c
@@ -218,6 +218,8 @@ find_my_exec(const char *argv0, char *retpath)
static int
resolve_symlinks(char *path)
{
+ // On GuixSD we *want* stuff relative to symlinks.
+ return 0;
#ifdef HAVE_READLINK
struct stat buf;
charorig_wd[MAXPGPATH],
--
2.18.0
***
Based on Julien's code comment, I'm guessing this somehow magically
ensures that the postgres executable is run from the within the
union-build directory where its symlink exists rather than from the
postgresql-13 directory where the symlink is pointing.
So now we appear to have a solution to the missing postgis control file
issue. However, I have a few comments still:
1. The postgresql-11 and postgresql-9.6 packages in
/gnu/packages/databases.scm are also missing Julien's symlink patch
and thus do not work in combination with postgis currently.
2. The postgis pa
Touchpad not working on Acer Aspire One laptop
Hi all, I successfully installed Guix on a Acer Aspire One laptop but the touchpad not working. Do you have any idea how I can make it work. This model has a SynPS/2 Synaptics TouchPad. Thanks for your help in advance! Best regards, Barnabas Beres
Re: guix system docker-image : missing root file system and 'incompatible bytecode kind'
Dear, On Sat, 31 Oct 2020 at 15:31, e...@beaver-labs.com wrote: > What I don't understand is that if I use `guix pack` I can get a docker > image with my software installed. However I must then run > docker run denatting:latest denat > to run the server. Passing the option ’--entry-point=denat’ to “guix pack -f docker” does not do the trick? > My goal here is to create a shepherd service so that the client just has > to run > docker run denatting:latest > without habing to specify a command, and with shepherd monitoring the > server and restarting it when it crashes. I have not looked at your config file. 1. Does your ’denat’ service simply work on a plain Guix? 2. Is your config correct? Because this is weird: --8<---cut here---start->8--- failed to start service 'file-systems' [...] failed to start service 'mcron' --8<---cut here---end--->8--- All the best, simon
