glib-networking: clean up & add installed tests #67957

jtojnar · 2019-09-02T17:20:14Z

built glib-networking
ran glib-networking.tests
checked that nothing relies on the dependencies formerly propagated (nix-review?)

auto_features are now enabled by default.

* Add maintainers * Add homepage * Correct license as per https://gitlab.gnome.org/GNOME/glib-networking/commit/b9c0324376a5e23b5f8210b0dd76858206baef26

This was done from the beginning for no apparent reason: NixOS@cc6ecae

So that this can be loaded from programs not depending on gsettings-desktop-schemas. Currently, this patch is much uglier than it could be due to https://gitlab.gnome.org/GNOME/glib/issues/1884

pkgs/development/libraries/libproxy/default.nix

worldofpeace · 2019-09-03T01:54:09Z

pkgs/development/libraries/glib-networking/default.nix

+    find "$installedTests/libexec" -type f -executable -print0 \
+      | while IFS= read -r -d "" file; do
+      echo "Wrapping program '$file'"
+      wrapProgram "$file" --prefix GIO_EXTRA_MODULES : "$out/lib/gio/modules"


Why just the gio modules? Couldn't we use wrapGAppsHook, disable it, and wrap them with its args?

I do not want to introduce dependencies on GTK, librsvg etc.

Right, can we make note of that so someone doesn't refactor it to this down the line?

Can we fix the assumptions about wrapGAppsHook instead? It even has apps in the name.

I really need to finish #43150

I really need to finish #43150

I do have notes locally on some things I wanted to add to #43150 but documentation writing escapes me also.
Though it would be good, just even as a start, to merge the parts about wrapGAppsHook which I believe was your original intent with this.

Can we fix the assumptions about wrapGAppsHook instead? It even has apps in the name.

When you say "it has apps in the name", it's confusing when it would be used for executables in a library?

I do have notes locally on some things I wanted to add to #43150 but documentation writing escapes me also.

I guess I will try to work on I today.

Though it would be good, just even as a start, to merge the parts about wrapGAppsHook which I believe was your original intent with this.

Yeah keeping the scope narrow will make it easier to consider it mergeable. It is just bad that the wrapper connects so much stuff.

When you say "it has apps in the name", it's confusing when it would be used for executables in a library?

For me, an app is primarily a graphical program that a person can use to carry out some task. Though there can be console apps for which we would not want to depend on GTK either. I would not consider such helper of library to be an app.

I do have notes locally on some things I wanted to add to #43150 but documentation writing escapes me also.

I guess I will try to work on I today.

Though it would be good, just even as a start, to merge the parts about wrapGAppsHook which I believe was your original intent with this.

Yeah keeping the scope narrow will make it easier to consider it mergeable. It is just bad that the wrapper connects so much stuff.

Great, I'd be happy to follow up on it once it's merged. And our roadmap can include what needs to be documented in the future.

worldofpeace · 2019-09-03T02:24:01Z

Huh, why does libsoup have?

nixpkgs/pkgs/development/libraries/libsoup/default.nix

Line 34 in 0c52651

propagatedUserEnvPackages = [ glib-networking.out ];

(in passthru)

worldofpeace · 2019-09-03T02:25:34Z

checked that nothing relies on the dependencies formerly propagated (nix-review?)

Hmm, maybe we can use staging-next for this 😄
Thinking it's low likelihood.

Also, I think glib-networking has the glib-pacrunner. Could that do with any wrapping?

jtojnar · 2019-09-03T02:46:22Z

Huh, why does libsoup have?

nixpkgs/pkgs/development/libraries/libsoup/default.nix

Line 34 in 0c52651

propagatedUserEnvPackages = [ glib-networking.out ];

(in passthru)

I suspect that is because we cannot add wrappers to libraries and we wanted to add TLS support to libsoup.

jtojnar · 2019-09-03T02:47:18Z

checked that nothing relies on the dependencies formerly propagated (nix-review?)

Hmm, maybe we can use staging-next for this smile
Thinking it's low likelihood.

Yeah looking through the reverse dependencies there do not seem to be any obvious issues.

jtojnar · 2019-09-03T13:53:45Z

Also, I think glib-networking has the glib-pacrunner. Could that do with any wrapping?

They use g_proxy_resolver_lookup_async so apparently, yes.

We no longer propagate dependencies so the comment is not relevant.

Add GSettings schemas required for GNOME helper.

worldofpeace

Builds and tests passes.

jtojnar · 2019-09-04T02:53:06Z

By the way, I used Coccinelle to create a base patch here as well:

spatch --sp-file hardcode-gsettings.cocci --dir proxy/ --in-place

@r1@
expression GSETTINGS_SCHEMA;
expression settings;
@@
- settings = g_settings_new (GSETTINGS_SCHEMA);
+ {
+ GSettingsSchemaSource *schema_source;
+ GSettingsSchema *schema;
+ schema_source = g_settings_schema_source_new_from_directory ("@gds_gsettings_path@",  g_settings_schema_source_get_default (), TRUE, NULL);
+ schema = g_settings_schema_source_lookup (schema_source, GSETTINGS_SCHEMA, FALSE);
+ settings = g_settings_new_full (schema, NULL, NULL);
+ g_settings_schema_source_unref (schema_source);
+ g_settings_schema_unref (schema);
+ }

Due to this GLib bug, I had to use a second patch:

@r2@
expression CHILD_SCHEMA;
expression settings;
expression parent_settings;
@@
- settings = g_settings_get_child (parent_settings, CHILD_SCHEMA);
+ {
+ GSettingsSchemaSource *schema_source;
+ GSettingsSchema *schema;
+ schema_source = g_settings_schema_source_new_from_directory ("@gds_gsettings_path@", g_settings_schema_source_get_default (), TRUE, NULL);
+ schema = g_settings_schema_source_lookup (schema_source, CHILD_SCHEMA, FALSE);
+ settings = g_settings_new_full (schema, NULL, NULL);
+ g_settings_schema_source_unref (schema_source);
+ g_settings_schema_unref (schema);
+ }

and then had to modify the files with sed -i 's/GNOME_PROXY_.*_CHILD_SCHEMA,/GNOME_PROXY_SETTINGS_SCHEMA "." \0/g' hardcode-gsettings.patch because Coccinelle complained about parse error when I tried to build the child schema id directly in the cocci patch.

jtojnar added 5 commits September 2, 2019 18:57

glib-networking: format with nixpkgs-fmt

f61b3e0

glib-networking: drop unnecessary flag

84f318e

auto_features are now enabled by default.

glib-networking: improve meta

8c7be1b

* Add maintainers * Add homepage * Correct license as per https://gitlab.gnome.org/GNOME/glib-networking/commit/b9c0324376a5e23b5f8210b0dd76858206baef26

glib-networking: stop propagating dependencies

cb1cc75

This was done from the beginning for no apparent reason: NixOS@cc6ecae

glib-networking: clean up

897a09f

jtojnar requested review from adisbladis, alyssais, basvandijk, edolstra, edwtjo, Ericson2314, FRidh, joachifm, LnL7, matthewbauer, Mic92, nbp, peti, Profpatsch, thoughtpolice, ttuegel and zimbatm as code owners September 2, 2019 17:20

ofborg bot added 6.topic: emacs 6.topic: erlang 6.topic: fetch 6.topic: GNOME GNOME desktop environment and its underlying platform 6.topic: golang 6.topic: haskell 6.topic: lua 6.topic: nixos labels Sep 2, 2019

ofborg bot removed 8.has: documentation 8.has: module (update) 2.status: merge conflict labels Sep 2, 2019

ofborg bot requested review from worldofpeace and hedning September 2, 2019 20:56

ofborg bot added 11.by: package-maintainer 10.rebuild-darwin: 11-100 10.rebuild-linux: 101-500 labels Sep 2, 2019

glib-networking: hardcode GSettings path to gnomeproxy module

ce40361

So that this can be loaded from programs not depending on gsettings-desktop-schemas. Currently, this patch is much uglier than it could be due to https://gitlab.gnome.org/GNOME/glib/issues/1884

jtojnar force-pushed the glib-networking-cleanup branch from 60dbad9 to d531045 Compare September 3, 2019 00:49

worldofpeace reviewed Sep 3, 2019

View reviewed changes

pkgs/development/libraries/libproxy/default.nix Show resolved Hide resolved

jtojnar mentioned this pull request Sep 3, 2019

libgdata: 0.17.10 → 0.17.11 #67939

Merged

6 tasks

worldofpeace reviewed Sep 3, 2019

View reviewed changes

jtojnar added 5 commits September 3, 2019 16:11

glib-networking: add installed tests

ecf6f4a

libproxy: remove outdated comment

a01a854

We no longer propagate dependencies so the comment is not relevant.

libproxy: split python outputs

d0ddf64

libproxy: format with nixpkgs-fmt

e9953b6

libproxy: wrap GSetings helper

0aa934a

Add GSettings schemas required for GNOME helper.

jtojnar force-pushed the glib-networking-cleanup branch from d531045 to 0aa934a Compare September 3, 2019 14:11

ofborg bot requested a review from worldofpeace September 3, 2019 14:30

worldofpeace approved these changes Sep 3, 2019

View reviewed changes

worldofpeace merged commit 42f63ff into NixOS:staging Sep 3, 2019

jtojnar deleted the glib-networking-cleanup branch September 3, 2019 14:55

ferdnyc mentioned this pull request Jun 1, 2022

meson: remove deprecated functions GSConnect/gnome-shell-extension-gsconnect#1370

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

glib-networking: clean up & add installed tests #67957

glib-networking: clean up & add installed tests #67957

jtojnar commented Sep 2, 2019 •

edited

worldofpeace Sep 3, 2019

jtojnar Sep 3, 2019

worldofpeace Sep 3, 2019

jtojnar Sep 3, 2019 •

edited

worldofpeace Sep 3, 2019 •

edited

jtojnar Sep 3, 2019

worldofpeace Sep 3, 2019

worldofpeace commented Sep 3, 2019

worldofpeace commented Sep 3, 2019 •

edited

jtojnar commented Sep 3, 2019

jtojnar commented Sep 3, 2019

jtojnar commented Sep 3, 2019

worldofpeace left a comment

jtojnar commented Sep 4, 2019

glib-networking: clean up & add installed tests #67957

glib-networking: clean up & add installed tests #67957

Conversation

jtojnar commented Sep 2, 2019 • edited

worldofpeace Sep 3, 2019

Choose a reason for hiding this comment

jtojnar Sep 3, 2019

Choose a reason for hiding this comment

worldofpeace Sep 3, 2019

Choose a reason for hiding this comment

jtojnar Sep 3, 2019 • edited

Choose a reason for hiding this comment

worldofpeace Sep 3, 2019 • edited

Choose a reason for hiding this comment

jtojnar Sep 3, 2019

Choose a reason for hiding this comment

worldofpeace Sep 3, 2019

Choose a reason for hiding this comment

worldofpeace commented Sep 3, 2019

worldofpeace commented Sep 3, 2019 • edited

jtojnar commented Sep 3, 2019

jtojnar commented Sep 3, 2019

jtojnar commented Sep 3, 2019

worldofpeace left a comment

Choose a reason for hiding this comment

jtojnar commented Sep 4, 2019

jtojnar commented Sep 2, 2019 •

edited

jtojnar Sep 3, 2019 •

edited

worldofpeace Sep 3, 2019 •

edited

worldofpeace commented Sep 3, 2019 •

edited