Add systemd-repart backend by vjr · Pull Request #959 · elementary/installer

vjr · 2026-10-02T03:02:09Z

Fixes #941

EDIT: See #959 (comment) for as-on-date work-in-progress screenshots.

vjr · 2026-10-02T03:03:36Z

 option('supported_languages', type : 'string', value : 'aa;ab;ae;af;ak;am;an;ar;as;ast;av;ay;az;ba;be;bg;bh;bi;bm;bn;bo;br;bs;ca;ce;ch;ckb;co;cr;cs;cu;cv;cy;da;de;de_CH;de_DE;dv;dz;ee;el;en;en_US;en_AU;en_CA;en_GB;eo;es;et;eu;fa;ff;fi;fj;fo;fr;fr_BE;fr_CA;fr_FR;fy;ga;gd;gl;gn;gu;gv;ha;he;hi;ho;hr;ht;hu;hy;hz;ia;id;ie;ig;ii;ik;io;is;it;iu;ja;jv;ka;kab;kg;ki;kj;kk;kl;km;kn;ko;kr;ks;ku;kv;kw;ky;la;lb;lg;li;ln;lo;lt;lu;lv;mg;mh;mi;mk;ml;mn;mo;mr;ms;mt;my;na;nb;nd;ne;ng;nl;nn;nr;nv;ny;oc;oj;om;or;os;pa;pi;pl;ps;pt;pt_PT;pt_BR;qu;rm;rn;ro;ru;rue;rw;sa;sc;sd;se;sg;si;sk;sl;sm;sma;sn;so;sq;sr;ss;st;su;sv;sw;szl;ta;te;tg;th;ti;tk;tl;tn;to;tr;ts;tt;tw;ty;udm;ug;uk;ur;uz;ve;vi;vo;wa;wo;xh;yi;yo;za;zh;zh_CN;zh_HK;zh_TW;zu', description : 'The list of supported languages')
 option('preferred_languages', type : 'string', value : 'en;zh;es;fr;pt;ru;de', description : 'A prioritized list of languages')
-option('installer_backend', type : 'combo', choices : ['distinst'], value : 'distinst')
+option('installer_backend', type : 'combo', choices : ['distinst', 'repart'], value : 'repart')


@danirabbit Do we want to set the default to repart at this time or leave it as distinst? I prefer to switch to repart.

Tbh I don't think this is the way to go. We can detect this at runtime no? We'd have to build two packages this way if we want to have the classic install image.

Can we do the runtime thing in a later PR? that will require UI changes to let the user pick which mode they want?

Yeah, both options should be in the installer.

The iso contains both images

Oh for some reason I thought it was two different images. I'm unsure how best to handle that 🤔

Latest commits (very much still experimental WIP branch) enable runtime backend selection, please take a look when you can. I'm sure there's a cleaner more elegant way to do all this?

vjr · 2026-10-02T03:25:52Z

@danirabbit i've addressed your feedback.

@jumpyvi i know both images are in the new ISO and i agree we should remove the compile time backend option and make it runtime user-selectable, but that can be in another PR, especially as I likely will fumble the UI changes lol.

I believe I've addressed everything for now.

danirabbit · 2026-10-02T03:36:16Z

It also seems like you didn't really try very hard to not just run everything directly as Terminal commands when we have for example GLib functions

vjr · 2026-10-02T03:47:24Z

It also seems like you didn't really try very hard to not just run everything directly as Terminal commands when we have for example GLib functions

You're right, it was a quick/lazy port from the bash script, I'll update the backend to use native stuff as much as available, thanks!

Co-authored-by: Vishal Rao <vishalrao@gmail.com>

Addressed.

vjr · 2026-10-08T08:22:06Z

@danirabbit @jumpyvi @elementary/desktop-developers @elementary/ux if you could provide review feedback, this PR is close to what I can do, next is to break this up into separate PRs.

danirabbit · 2026-10-13T17:10:14Z


    private void load_partitioning_view () {
-        var partitioning_view = new PartitioningView (MINIMUM_SPACE);
+        var partitioning_view = new PartitioningView (Daemon.get_default ().get_backend () == REPART ? MINIMUM_SPACE_MODERN : MINIMUM_SPACE);


Seems like the backend should return the correct value instead of having those values live here?

danirabbit · 2026-10-13T17:15:54Z

+
+[DBus (name = "io.elementary.InstallerDaemon")]
+public class InstallerDaemon.BackendProxy : GLib.Object {
+    private InstallerDaemon.Backend backend = InstallerDaemon.Backend.UNKNOWN;


get; private set instead so we can get right of the get_backend func?

danirabbit · 2026-10-13T17:16:31Z

+            case InstallerDaemon.Backend.DISTINST:
+                backend_proxy = new InstallerDaemon.DistinstBackend ();
+                break;
+            case InstallerDaemon.Backend.REPART:


You can use implicit namespace here

Suggested change

case InstallerDaemon.Backend.DISTINST:

backend_proxy = new InstallerDaemon.DistinstBackend ();

break;

case InstallerDaemon.Backend.REPART:

case DISTINST:

backend_proxy = new InstallerDaemon.DistinstBackend ();

break;

case REPART:

danirabbit · 2026-10-13T17:18:16Z

+    private const string SQUASH_MOUNT = "/mnt/source-image";
+
+    private void log_message (InstallerDaemon.LogLevel level, string format, ...) {
+        var msg= format.vprintf (va_list ());


Suggested change

var msg= format.vprintf (va_list ());

var msg = format.vprintf (va_list ());

danirabbit · 2026-10-13T17:19:39Z

+    public bool has_tpm2 () throws GLib.Error {
+        try {
+            var output = run_capture ({"systemd-analyze", "has-tpm2"});
+            var lines = output.split ("\n");
+            return lines.length > 0 && lines[0].strip () == "yes";
+        } catch (GLib.Error e) {
+            log_message (InstallerDaemon.LogLevel.WARN, "TPM2 check failed: %s", e.message);
+            return false;
+        }
+    }


Isn't systemd-analyze available regardless of backend?

danirabbit · 2026-10-13T17:22:39Z

+    public abstract bool has_tpm2 () throws GLib.Error;
+    protected bool default_has_tpm2 () {
+        return false;
+    }


I think this is maybe not entirely telling the truth. There's 2 things here, one is whether the system has tpm2 (which presumably we can always check with systemd-analyze regardless of backend). The other is if the backend supports tpm2 (which should be false by default and then the backend can overwrite it).

Then we can early return in has_tpm2 () if !Backend.supports_tpm ()

danirabbit · 2026-10-13T17:25:00Z

        }

-        var apt_sources = File.new_for_path ("/etc/apt/sources.list.d/elementary.list");
+        var apt_sources = File.new_for_path ("/etc/apt/sources.list.d/elementary.sources");


Why have you changed this?

danirabbit · 2026-10-13T17:26:50Z

+    private Gtk.CheckButton use_tpm_checkbutton;

    private const string SKIP_STRING = _("Don’t Encrypt");
+    private const string TPM_STRING = _("Encrypt Using TPM");


Is there a reason why we wouldn't want to use TPM if it's available? Does this even need to be an option?

If it does need to be an option, we need to provide information that helps people decide which choice is right for them

danirabbit · 2026-10-13T17:29:01Z

@@ -1,3 +1,2 @@
 option('supported_languages', type : 'string', value : 'aa;ab;ae;af;ak;am;an;ar;as;ast;av;ay;az;ba;be;bg;bh;bi;bm;bn;bo;br;bs;ca;ce;ch;ckb;co;cr;cs;cu;cv;cy;da;de;de_CH;de_DE;dv;dz;ee;el;en;en_US;en_AU;en_CA;en_GB;eo;es;et;eu;fa;ff;fi;fj;fo;fr;fr_BE;fr_CA;fr_FR;fy;ga;gd;gl;gn;gu;gv;ha;he;hi;ho;hr;ht;hu;hy;hz;ia;id;ie;ig;ii;ik;io;is;it;iu;ja;jv;ka;kab;kg;ki;kj;kk;kl;km;kn;ko;kr;ks;ku;kv;kw;ky;la;lb;lg;li;ln;lo;lt;lu;lv;mg;mh;mi;mk;ml;mn;mo;mr;ms;mt;my;na;nb;nd;ne;ng;nl;nn;nr;nv;ny;oc;oj;om;or;os;pa;pi;pl;ps;pt;pt_PT;pt_BR;qu;rm;rn;ro;ru;rue;rw;sa;sc;sd;se;sg;si;sk;sl;sm;sma;sn;so;sq;sr;ss;st;su;sv;sw;szl;ta;te;tg;th;ti;tk;tl;tn;to;tr;ts;tt;tw;ty;udm;ug;uk;ur;uz;ve;vi;vo;wa;wo;xh;yi;yo;za;zh;zh_CN;zh_HK;zh_TW;zu', description : 'The list of supported languages')
 option('preferred_languages', type : 'string', value : 'en;zh;es;fr;pt;ru;de', description : 'A prioritized list of languages')
-option('installer_backend', type : 'combo', choices : ['distinst'], value : 'distinst')


I think we should still keep this because there was talk about replacing distinst with calamares

jumpyvi · 2026-10-19T17:56:25Z

Could distinst be optional so it can be built only with repart support?

Distinst is not fun to build

danirabbit · 2026-10-19T20:15:53Z

    private const int DBUS_TIMEOUT_MSEC = 600 * 1000;

    [DBus (name = "io.elementary.InstallerDaemon")]
    private interface InstallerInterface : GLib.DBusProxy {


We should probably move this to common instead of duplicating it

Add systemd-repart backend

2b7fcfa

vjr requested review from a team and danirabbit October 2, 2026 03:02

vjr self-assigned this Oct 2, 2026

vjr commented Oct 2, 2026

View reviewed changes

This comment was marked as resolved.

Sign in to view

vjr added 3 commits October 2, 2026 08:45

Use SPDX license format

c0647bf

Add line breaks

47ae752

Fix some logging and error conditions

bbbd4dd

vjr requested review from danirabbit and jumpyvi October 2, 2026 03:25