dia: 2017-06-22 → 2020-04-07 #84555

jtojnar · 2020-04-07T06:26:23Z

Motivation for this change

Getting rid of GNOME 2 package set #39976

cc @viric

Things done

[ x] Tested using sandboxing (nix.useSandbox on NixOS, or option sandbox in nix.conf on non-NixOS linux)
Built on platform(s)
- NixOS
- macOS
- other Linux distributions
Tested via one or more NixOS test(s) if existing and applicable for the change (look inside nixos/tests)
Tested compilation of all pkgs that depend on this change using nix-shell -p nixpkgs-review --run "nixpkgs-review wip"
Tested execution of all binary files (usually in ./result/bin/)
Determined the impact on package closure size (by running nix path-info -S before and after)
Ensured that relevant documentation is up to date
Fits CONTRIBUTING.md.

and reorder the attributes

veprbl · 2020-04-07T08:40:43Z

On darwin this fails with

2020-04-07 04:35:09.971 dia[70685:3631021] *** WARNING: Method userSpaceScaleFactor in class NSView is deprecated on 10.7 and later. It should not be used in new applications. Use convertRectToBacking: instead. 

** (dia:70685): WARNING **: 04:35:10.471: Unable to find object type: GRAFCET - Step

** (dia:70685): WARNING **: 04:35:10.471: Unable to find object type: GRAFCET - Step

<many similar warnings>

** (dia:70685): WARNING **: 04:35:10.564: Unable to find object type: Flowchart - Predefined Process

** (dia:70685): WARNING **: 04:35:10.564: Unable to find object type: Standard - Box

** (dia:70685): CRITICAL **: 04:35:10.574: Couldn't find standard objects when looking for object-libs in '/nix/store/jbwri07nqs0qcybyv576rfr12id4n08v-dia-unstable-2020-04-07/lib/dia/'; exiting...

I came up with a following workaround:

diff --git a/pkgs/applications/graphics/dia/default.nix b/pkgs/applications/graphics/dia/default.nix
index 6eee16145a7..5acba6f5895 100644
--- a/pkgs/applications/graphics/dia/default.nix
+++ b/pkgs/applications/graphics/dia/default.nix
@@ -58,6 +58,12 @@ stdenv.mkDerivation {
       build-aux/post-install.py
   '';
 
+  postInstall = stdenv.lib.optionalString stdenv.isDarwin ''
+    for dylib in "$out"/lib/dia/*.dylib; do
+      ln "''${dylib}" "''${dylib%.*}.so"
+    done
+  '';
+
   meta = with stdenv.lib; {
     description = "GNOME Diagram drawing software";
     homepage = "https://wiki.gnome.org/Apps/Dia";

After that dia launches it is more or less usable, except for some rendering problems:

jtojnar · 2020-04-07T11:22:07Z

@veprbl pushed a new revision, re-adding gtk-mac-integration-gtk2 and merging https://gitlab.gnome.org/GNOME/dia/-/merge_requests/68. Does that fix it for you?

veprbl · 2020-04-07T18:31:31Z

@jtojnar Thank you for implementing the fix. It does what needs to be done, but misses some of the libraries:

$ ls -1 /nix/store/jaq7jss0hmlglpsjb4w7bjgp9g7y5rpx-dia-unstable-2020-04-07/lib/dia/
libaadl_objects.dylib
libcairo_filter.so
libcgm_filter.so
libchronogram_objects.dylib
libcustom_lines_objects.dylib
libcustom_objects.dylib
libdb_objects.dylib
libdxf_filter.so
liber_objects.dylib
libflowchart_objects.dylib
libfs_objects.dylib
libgrafcet_objects.dylib
libhpgl_filter.so
libistar_objects.dylib
libjackson_objects.dylib
libkaos_objects.dylib
liblayout_filter.so
libmetapost_filter.so
libmisc_objects.dylib
libnetwork_objects.dylib
libpdf_filter.so
libpgf_filter.so
libpixbuf_filter.so
libpostscript_filter.so
libpstricks_filter.so
libpython_plugin.so
libsadt_objects.dylib
libshape_filter.so
libstandard_objects.dylib
libsvg_filter.so
libuml_objects.dylib
libvdx_filter.so
libwpg_filter.so
libxfig_filter.so
libxslt_filter.dylib

Regarding the rendering issues, I don't see much change. I've played with a window a bit more, I've noticed it blanks the window on resize, but depending on the size of the window three different things can happen. If the window is "smaller", it will render window as black (which macOS seems to screenshot as transparent color):

If it is "bigger" it actually renders the proper grey background and able to render most of the icons:

And, finally, excessively small windows will be black, but will turn to white garbled texture on mouse hover (see my first screenshot for example).

Most buts get redrawn on mouse hover, so this is almost usable.

veprbl · 2020-04-07T18:40:56Z

pkgs/applications/graphics/dia/default.nix

+    zlib
+    cairo
+  ] ++ stdenv.lib.optionals stdenv.isDarwin [
+    gtk-mac-integration-gtk2


Doesn't seem like this is used:

# grep -ri gtk_mac src/ # grep -ri gtk-mac src/ #

I would expect it to be somehow magically linked. Is it not listed in nix-store --query --tree or nix why-depends?

# nix-store --query --tree /nix/store/qflajk99al4hk1893gz6z2m996x6l930-dia-0.97.3.20170622 | grep mac +---/nix/store/brvfal9n3wi562r98q1rgicxcjlyh7i1-gtk-mac-integration-2.0.8 | +---/nix/store/brvfal9n3wi562r98q1rgicxcjlyh7i1-gtk-mac-integration-2.0.8 [...] # nix-store --query --tree /nix/store/jaq7jss0hmlglpsjb4w7bjgp9g7y5rpx-dia-unstable-2020-04-07 | grep mac #

Hmm, I can still grep GtkosxApplication in the source code. Looks like HAVE_MAC_INTEGRATION was not ported to Meson.

Strangely enough, old dia also doesn't have the bug:

And it does use the system menubar, like it's supposed to do with gtk-mac-integration.

Does the latest change help?

@jtojnar The build is now failing with

../app/interface.c:1060:34: error: use of undeclared identifier 'dia_app_icon' gdk_pixbuf_new_from_inline (-1, dia_app_icon, FALSE, NULL)); ^ 1 error generated.

After applying

diff --git a/app/interface.c b/app/interface.c --- a/app/interface.c +++ b/app/interface.c @@ -1057,7 +1057,7 @@ _create_mac_integration (GtkWidget *menubar) gtk_widget_hide (menubar); /* not working, it's shown elsewhere */ /* setup the dock icon */ gtkosx_application_set_dock_icon_pixbuf (theOsxApp, - gdk_pixbuf_new_from_inline (-1, dia_app_icon, FALSE, NULL)); + pixbuf_from_resource ("/org/gnome/Dia/icons/org.gnome.Dia.png")); } /* Don't quit without asking to save files first */ g_signal_connect (theOsxApp, "NSApplicationBlockTermination",

I get it to compile and run. I now get a proper dia icon in the dock and system-wide menu works again. There is no improvement with rendering.

veprbl · 2020-04-07T20:26:49Z

Just tested 2889df2a66e, it starts up fine.

edit: no dylibs in $out/lib/dia anymore

ZanderBrown · 2020-04-07T21:10:27Z

Just checking: Your shipping this as unstable/testing/devel right?

jtojnar · 2020-04-07T23:35:11Z

@ZanderBrown Yup, we use dia-unstable as the package name.

jtojnar · 2020-04-08T00:05:23Z

@veprbl Would you be willing to bisect the graphical glitches? Should take roughly eight steps.

(You would run git bisect start master b86085dfe2b048a2d37d587adf8ceba6fb8bc43c in the Dia repo, change the nixpkgs expression’s src to your dia checkout, and then build it and try to run it and mark the bisection step with git bisect good/bad.)

* Upgrade three years old git snapshot to a fresh one * Switches to Meson * Drops GNOME 2 dependencies

veprbl · 2020-04-08T13:58:18Z

@jtojnar
The bisect didn't go as smooth as planned. Many builds failed for various reasons and there was some bug that caused dia to crash on start. I patched couple of the bugs on the go and arrived to the following expression:

with import <nixpkgs> {};

stdenv.mkDerivation {
  name = "dia";

  src = lib.cleanSource ./diagit;

  nativeBuildInputs = [
    meson
    ninja
    pkg-config
    gettext
    python3 # for install script
    gtk2 # for gtk-update-icon-cache
    #libxslt # for xsltproc
    desktop-file-utils
    intltool
  ];

  buildInputs = [
    glib
    gtk2
    libxml2
    python2
    zlib
    cairo
  ];

  postPatch = ''
    patchShebangs \
      generate_run_with_dia_env.sh \
      build-aux/post-install.py
    substituteInPlace plug-ins/wmf/wmf.cpp \
      --replace '#include "config.h"' ""
    substituteInPlace objects/standard/outline.c \
      --replace '#include "tool-icons.h"' ""
  '';

  postInstall = ''
    for dylib in "$out"/lib/dia/*.dylib; do
      ln "''${dylib}" "''${dylib%.*}.so"
    done
  '';
}

What I learned was that there is a "good" version of dia that is able to be built with this (or a similar) meson-based expression, which is close enough to what you propose in this PR.

edit: updated bisect result after fixing more irrelevant failures

There are only 'skip'ped commits left to test.
The first bad commit could be any of:
651f7eb4810de4f94f04a106b0a45eb0495b1381
fd5d59244fd92909896c640578a6fca940aef47f
642282d842de4e59a06c0d7b53418683352a6a65
ea21bcab9d8740cbf7feef81359525128734cea6
665f9429dc7ad9a27cc6c0c265e586f37a5db33a
c1ec976f1e450fc5cc9873b929db0bf3b96b90ae
659179878502e515b6748c3e76dd3d51422a1925
ff4c247dea72c436e0c3b206039eb7e72dd60c6c
1b7726cd9044852af07fa296e2722237ef1878ee
af3c7836c0bf9aa7ac62fbac9a3dd569a9c3db81
9927ab22db91508f594f1d31f365cd02631e6c28
d0137e84d8cb47feb873911fd49ae18ff180c9bc
33e5aaa3585f040898159bc16d1d7ca67523b0de
d0339fa23a511feb44e653655dd24b191bf6b899
5654328926b2b4872b4e6056c3bf51adf47b0fa2
5e148bac88bded7d9e21f34b2f30f24ae0c1e9eb
1bef8c839986ec6ca89a44061154c483e3a19281
We cannot bisect more!

jtojnar · 2020-04-08T14:08:09Z

Yeah, unfortunately, it is not easy to automate it. At best you will be able to keep automake & co. and preConfigure from the original expression and use the following hack to disable meson when not available:

  ${if builtins.pathExists "${src}/meson.build" then null else "dontUseMesonConfigure"} = true;
  ${if builtins.pathExists "${src}/meson.build" then null else "dontUseNinjaBuild"} = true;
  ${if builtins.pathExists "${src}/meson.build" then null else "dontUseNinjaInstall"} = true;

With 8 steps, it is also feasible, if annoying, to run the bisection manually.

veprbl · 2020-04-08T14:15:35Z

@jtojnar I think you got me wrong. The breakage happened after meson was introduced. I am now trying to bisect the crash to learn how to fix it and, maybe, narrow down the search.

veprbl · 2020-04-08T14:38:08Z

I guess, this has to do with new cairo renderer in dia.

ZanderBrown · 2020-04-08T15:09:03Z

Most likely, interestingly Windows & X11 haven't had an issue like that with rendering the canvas

However as of a few days ago line/arrow chooser in the properties dialogue do, except that code hasn't changed in months

It seems when rendering goes wrong in gtk2 it goes really wrong

jtojnar · 2020-04-08T15:10:55Z

The question is do we ship this, or wait for a fix (possibly GTK 3 port)?

veprbl · 2020-04-08T15:14:52Z

@jtojnar I would suggest we pin dia to a latest pre-cairo version for darwin. I can prepare a change for that.

ZanderBrown · 2020-04-08T15:15:31Z

This exact version? I'd say no, even as an unstable package that's way too broken

Of course I'm interested what's going on, especially as it's almost certainly my fault. Might try a VM at somepoint

veprbl · 2020-04-08T15:20:03Z

@ZanderBrown Let me know if there is anything I can try to help debugging this.

jtojnar · 2020-04-08T15:20:58Z

Okay, since it does not really improve that much and it is far from the worst gnome2 library usage offender 😿, we will table this PR for now.

ZanderBrown · 2020-04-08T15:23:57Z

far from the worst gnome2 library usage offender

What is it using? if it's libgnomeui there is a configure flag for that

jtojnar · 2020-04-08T15:25:34Z

libart_lgpl and libgnomeui

ZanderBrown · 2020-04-08T15:26:47Z

Ah well until we figure out the cairo issue libart_lgpl is sticking around, for libgnomeui just --disable it, isn't needed

stale · 2021-06-04T15:03:15Z

I marked this as stale due to inactivity. → More info

armeenm · 2022-04-25T15:56:03Z

Any updates on this?

jtojnar · 2022-04-25T20:58:01Z

Needs more upstream work: https://gitlab.gnome.org/GNOME/dia/-/merge_requests/87, https://gitlab.gnome.org/GNOME/dia/-/merge_requests/86, https://gitlab.gnome.org/GNOME/dia/-/issues/520

veprbl · 2023-01-08T01:07:32Z

Superseded by #209587

dia: format with nixpkgs-fmt

c06918f

and reorder the attributes

jtojnar added this to To do in Picking up garbage via automation Apr 7, 2020

jtojnar marked this pull request as ready for review April 7, 2020 06:26

jtojnar moved this from To do to In progress in Picking up garbage Apr 7, 2020

ofborg bot requested a review from 7c6f434c April 7, 2020 06:34

ofborg bot added 10.rebuild-darwin: 1-10 10.rebuild-darwin: 1 10.rebuild-linux: 1-10 10.rebuild-linux: 1 labels Apr 7, 2020

jtojnar force-pushed the dia-upgrade branch from 8de1911 to 9f52b11 Compare April 7, 2020 07:07

ofborg bot added the ofborg-internal-error label Apr 7, 2020

jtojnar force-pushed the dia-upgrade branch 2 times, most recently from 418ca5c to 8dbce49 Compare April 7, 2020 11:20

grahamc removed the ofborg-internal-error label Apr 7, 2020

veprbl reviewed Apr 7, 2020

View reviewed changes

jtojnar force-pushed the dia-upgrade branch from 8dbce49 to 2889df2 Compare April 7, 2020 20:16

jtojnar force-pushed the dia-upgrade branch from 2889df2 to 630bf06 Compare April 7, 2020 20:58

dia: 2017-06-22 → 2020-04-07

6b01789

* Upgrade three years old git snapshot to a fresh one * Switches to Meson * Drops GNOME 2 dependencies

jtojnar force-pushed the dia-upgrade branch from 630bf06 to 6b01789 Compare April 8, 2020 01:35

jtojnar mentioned this pull request Apr 8, 2020

gnome2 cleanups #84698

Merged

jtojnar added the 2.status: wait-for-upstream Waiting for upstream fix (or their other action). label Apr 8, 2020

jtojnar marked this pull request as draft May 25, 2020 12:39

ryantm added 2.status: merge conflict and removed 2.status: merge conflict labels Oct 3, 2020

veprbl mentioned this pull request Jan 30, 2021

dia: fix CVE-2019-19451 #111327

Merged

10 tasks

stale bot added the 2.status: stale https://github.com/NixOS/nixpkgs/blob/master/.github/STALE-BOT.md label Jun 4, 2021

stale bot removed the 2.status: stale https://github.com/NixOS/nixpkgs/blob/master/.github/STALE-BOT.md label Apr 25, 2022

stale bot added the 2.status: stale https://github.com/NixOS/nixpkgs/blob/master/.github/STALE-BOT.md label Oct 30, 2022

veprbl closed this Jan 8, 2023

Picking up garbage automation moved this from In progress to Done Jan 8, 2023

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

dia: 2017-06-22 → 2020-04-07 #84555

dia: 2017-06-22 → 2020-04-07 #84555

jtojnar commented Apr 7, 2020

veprbl commented Apr 7, 2020 •

edited

jtojnar commented Apr 7, 2020

veprbl commented Apr 7, 2020

veprbl Apr 7, 2020

jtojnar Apr 7, 2020

veprbl Apr 7, 2020

jtojnar Apr 7, 2020

veprbl Apr 7, 2020

jtojnar Apr 7, 2020

veprbl Apr 7, 2020 •

edited

veprbl commented Apr 7, 2020 •

edited

ZanderBrown commented Apr 7, 2020

jtojnar commented Apr 7, 2020

jtojnar commented Apr 8, 2020

veprbl commented Apr 8, 2020 •

edited

jtojnar commented Apr 8, 2020

veprbl commented Apr 8, 2020

veprbl commented Apr 8, 2020

ZanderBrown commented Apr 8, 2020

jtojnar commented Apr 8, 2020

veprbl commented Apr 8, 2020

ZanderBrown commented Apr 8, 2020

veprbl commented Apr 8, 2020

jtojnar commented Apr 8, 2020

ZanderBrown commented Apr 8, 2020

jtojnar commented Apr 8, 2020

ZanderBrown commented Apr 8, 2020

stale bot commented Jun 4, 2021

armeenm commented Apr 25, 2022

jtojnar commented Apr 25, 2022

veprbl commented Jan 8, 2023

dia: 2017-06-22 → 2020-04-07 #84555

dia: 2017-06-22 → 2020-04-07 #84555

Conversation

jtojnar commented Apr 7, 2020

Motivation for this change

Things done

veprbl commented Apr 7, 2020 • edited

jtojnar commented Apr 7, 2020

veprbl commented Apr 7, 2020

veprbl Apr 7, 2020

Choose a reason for hiding this comment

jtojnar Apr 7, 2020

Choose a reason for hiding this comment

veprbl Apr 7, 2020

Choose a reason for hiding this comment

jtojnar Apr 7, 2020

Choose a reason for hiding this comment

veprbl Apr 7, 2020

Choose a reason for hiding this comment

jtojnar Apr 7, 2020

Choose a reason for hiding this comment

veprbl Apr 7, 2020 • edited

Choose a reason for hiding this comment

veprbl commented Apr 7, 2020 • edited

ZanderBrown commented Apr 7, 2020

jtojnar commented Apr 7, 2020

jtojnar commented Apr 8, 2020

veprbl commented Apr 8, 2020 • edited

jtojnar commented Apr 8, 2020

veprbl commented Apr 8, 2020

veprbl commented Apr 8, 2020

ZanderBrown commented Apr 8, 2020

jtojnar commented Apr 8, 2020

veprbl commented Apr 8, 2020

ZanderBrown commented Apr 8, 2020

veprbl commented Apr 8, 2020

jtojnar commented Apr 8, 2020

ZanderBrown commented Apr 8, 2020

jtojnar commented Apr 8, 2020

ZanderBrown commented Apr 8, 2020

stale bot commented Jun 4, 2021

armeenm commented Apr 25, 2022

jtojnar commented Apr 25, 2022

veprbl commented Jan 8, 2023

veprbl commented Apr 7, 2020 •

edited

veprbl Apr 7, 2020 •

edited

veprbl commented Apr 7, 2020 •

edited

veprbl commented Apr 8, 2020 •

edited