Gajim - 2017-10-19


  1. concerto lovetox: re: http://paste.debian.net/991155/ > - I'm not automatically logged in or asked for my password (as usually happens) > - when I change status to Available to login, I just see my own JID (I do have the option for that enabled), and nobody else's. > lovetox: all menu entries in the Gajim menu are grayed out, too, so I can't disable plugins the usual way. ...any ideas? Gajim's become completely unusable because of this, and not even subsequent upgrades have fixed it. ._.
  2. Asterix Running with -v to see what happens would help
  3. concerto Asterix: cool, I'll try it
  4. lovetox concerto just go to your plugins folder and delete some plugins
  5. lovetox ~/.local/share/gajim/plugins
  6. zuglufttier lovetox, when you enter a message that is longer than one line, gajim will add a blank line below the current text. I don't think this is wanted ;)
  7. zuglufttier https://files.dismail.de/data/6bcce3d7-29c4-46f4-b2f1-71d125c5a85f/2017-10-19-110235_1488x1176_scrot.png
  8. lovetox its a gtk bug
  9. lovetox i reported it
  10. mimi89999 zuglufttier: I already reported it.
  11. lovetox until they fix it we have to live with it
  12. lovetox :)
  13. zuglufttier OK.
  14. mimi89999 lovetox: Appears to be fixed in the new ui version.
  15. zuglufttier Move over to QT :D
  16. lovetox it is not mimi89999
  17. lovetox at least for me :)
  18. mimi89999 Let me check again...
  19. lovetox you can try and run this test code
  20. lovetox https://bug788504.bugzilla-attachments.gnome.org/attachment.cgi?id=360943
  21. lovetox then report to that issue, that it also effects you
  22. lovetox https://bugzilla.gnome.org/show_bug.cgi?id=788504
  23. zuglufttier I commented on that bug.
  24. lovetox thanks
  25. Link Mauve I can reproduce this bug on all of the default themes, with the 3.22.24 version in ArchLinux.
  26. lovetox could you also please write a sentence into the bug ticket, the one dev can not reproduce it on his system, i want to show him that it effects many systems
  27. Link Mauve There is nothing special about ArchLinux, maybe just pinging other people on #gtk+ would help instead.
  28. Link Mauve lovetox, also seeing zuglufttier’s screenshot, Gajim desperately needs higher resolution icons.
  29. zuglufttier Oh yes!
  30. Link Mauve And also to not render avatars at a 1:1 resolution, to then upscale them again.
  31. zuglufttier Also: No icons for user state.
  32. Link Mauve See https://dev.gajim.org/gajim/gajim/issues/8709 and its dup.
  33. andrey.g I couldn't reproduce it on gtk2 based MATE with gtk3 package 3.22.21. But instead I've trapped into another bug. Since I'm not using native gtk3 theme, I had to open the integrated inspector (ctrl+shift+i) to switch to Adwaita for the reproducer. Then I've played with the TextView and then without to close the inspector, I've closed the main window and then the complete DE freezed.
  34. lovetox Link Mauve, from where do you get that we render avatars 1:1 and then upsacle it?
  35. lovetox and why are you sure its a upscale, 1:1 could mean 1000x1000 px
  36. zuglufttier Aren't vector graphics the way to go?
  37. mathieui lovetox, because the avatars on the screenshot are blurry
  38. mathieui even though they probably are higher resolution than this
  39. Link Mauve lovetox, I mean 1:1 of the size you plan on displaying them at.
  40. lovetox thats the same bug as with the icons
  41. Link Mauve zuglufttier, yes! See my avatar.
  42. lovetox we render them at a fixed pixel size
  43. lovetox not taking into account HDPI
  44. lovetox i dont exactly know how these hdpi screens work, but it seems if i set a icon to 16 px its then scaled up to double that or something like that
  45. lovetox so i first have to find out if a hdpi screen is used, then get the scaling factor then multiply the px with that factor
  46. lovetox for icons this is probably easier, becaue gtk could do this for us if we set all custom icons correctly up
  47. Link Mauve lovetox, basically, the logical pixel size is decorrelated from the actual rendered size.
  48. Link Mauve And it can change at runtime, for example when moving a window from one screen to another.
  49. Link Mauve The correct thing to do is to let the toolkit render as much as possible, but you may handle an event on the window changing its displayed pixel size and recompute avatars, or something.
  50. Link Mauve From a quick look at Glade, I’m not sure a GtkWindow has that.
  51. lovetox not the window itself
  52. lovetox https://lazka.github.io/pgi-docs/#Gdk-3.0/classes/Screen.html#Gdk.Screen
  53. lovetox hm yeah i will experiment with that should not be too hard
  54. lovetox configure
  55. lovetox https://lazka.github.io/pgi-docs/#Gdk-3.0/classes/Window.html#Gdk.Window.get_scale_factor
  56. Link Mauve Ah, configure.
  57. Link Mauve That’s what I was looking for.
  58. zuglufttier I use these settings in my .xinitrc: https://wiki.archlinux.org/index.php/HiDPI#GDK_3_.28GTK.2B_3.29
  59. lovetox zuglufttier, interested to test a patch?
  60. lovetox https://paste.gajim.org/view/73465f75
  61. lovetox this would just for the GC Roster avatars
  62. zuglufttier Sure, how do I apply the patch?
  63. lovetox hm search for groupchat_control.py on your computer
  64. lovetox and remove the line with the "-" and add the lines with the "+" around line 1614
  65. lovetox or comment the one line instead of removing so you can reverse it easily
  66. zuglufttier lovetox: 10/19/2017 13:12:43 (E) gajim.c.ged Error while running an even handler: <bound method GroupchatControl._nec_gc_presence_received of <gajim.groupchat_control.GroupchatControl object at 0x7f5b5da5e3c8>> Traceback (most recent call last): File "/home/zlt/git/gajim/gajim/common/ged.py", line 93, in raise_event if handler(*args, **kwargs): File "/home/zlt/git/gajim/gajim/groupchat_control.py", line 1825, in _nec_gc_presence_received affiliation, obj.status, obj.real_jid, obj.avatar_sha) File "/home/zlt/git/gajim/gajim/groupchat_control.py", line 1944, in add_contact_to_roster self.draw_avatar(gc_contact) File "/home/zlt/git/gajim/gajim/groupchat_control.py", line 1615, in draw_avatar factor = self.parent_win.window.get_scale_factor() AttributeError: 'NoneType' object has no attribute 'window'
  67. Asterix hehe minimized GC I guess ...
  68. lovetox that lead to so many bug ^^
  69. lovetox hm then this is more complicated, i didnt know we draw avatars even if there is no window
  70. zuglufttier Oh well, I can live with that for now ;)
  71. zuglufttier But blurry graphics can be really annoying.
  72. lovetox its always hard to develop for something if you cant test it
  73. zuglufttier I know... You could just try the settings from the arch wiki on your device.
  74. zuglufttier It will make all your applications very big but you also get to see the problems.
  75. zuglufttier You don't need any real hardware for that, it's just software settings.
  76. lovetox ok yeah i see this works
  77. zuglufttier Enjoy big and beautiful fonts :D
  78. zuglufttier Experience the terror of scaling.
  79. Link Mauve Scaling in GTK+ actually works very well.
  80. Link Mauve I can’t say the same about any other toolkit…
  81. Holger Link Mauve: Hah! I was looking for someone who could tell me how to scale GTK+ UIs for my high resolution :-)
  82. Link Mauve Holger, use Wayland.
  83. Holger Outside GNOME, that is. Inside GNOME it just magically looks good. But so far I failed tricking the processes to believe they're running under GNOME.
  84. Holger Link Mauve: Sigh.
  85. Link Mauve Holger, which DE are you using?
  86. Holger I'm using the i3 window manager.
  87. zuglufttier Yeah, using gnome might be fine :D I'm using dwm. GTK 3 always looks a bit broken.
  88. Holger Holger: But the notebook in question is at home while I'm at work, so I can't test things right now anyway. But I'd be highly interested in a solution.
  89. Holger Er, Link Mauve.
  90. Holger I mostly fixed GTK+2 programs via the gtkrc. But I found no place to adjust things for GTK+3.
  91. Holger (Same issue with Qt 5, e.g. Swift is unusable.)
  92. Link Mauve Holger, there is a reimplementation of i3 called Sway, which should be mostly compatible.
  93. Holger Yes I've seen it. Wayland is not really an option for me right now.
  94. Link Mauve X11 is fundamentally broken, it doesn’t have any concept of multiple outputs, nor of intrinsic scaling factor.
  95. Holger I have a NetBSD box; and we have no Wayland at work; and I'd like to keep my setups as similar as possible.
  96. Holger Link Mauve: Multiple outputs work just fine in practice though.
  97. Holger Link Mauve: So scaling should be X11's/Wayland's job? Wasn't your point that this is a toolkit thing?
  98. Holger If I did global scaling I'd have to fix all non-GTK/Qt stuff, because that's all fine right now :-) But that might work.
  99. Link Mauve Holger, no, scaling is the toolkit’s job, mostly.
  100. Link Mauve But toolkits can’t work with missing information, about that.
  101. Holger What do they need besides resolution/DPI?
  102. zuglufttier Holger, QT 5 can do scaling pretty nicely: https://wiki.archlinux.org/index.php/HiDPI#GUI_toolkits
  103. Holger zuglufttier: I played with that and didn't achieve a usable result.
  104. Link Mauve On Wayland, any window knows the scaling factor of each output, the outputs it is on currently, and it can specify at which scaling factor the content has been drawn; the compositor can then upscale/downscale the buffer if there are outputs on which the window is displayed which have a mismatched scaling factor.
  105. Holger zuglufttier: Whatever scaling you apply, you don't end up with UI elements that have a size that's (a) reasonable and (b) large enough to display the text in that element.
  106. Link Mauve A simple client, such as any Xwayland client, will render at scaling factor 1 no matter what, so the compositor will make it look blurry but at a correct resolution on any HiDPI output.
  107. Link Mauve A better client will react on entering a scale=2 output by resizing its buffer and specify it’s now scale=2, so the compositor will draw it with 1:1 pixels on that output but with 0.5:1 pixels on any scale=1 output it may still be spanning on.
  108. Holger Ah, so the point is display-specific scaling if you have multiple displays?
  109. Holger I get that, but I think that's not my problem.
  110. Link Mauve That, and in general to make clients aware of which scaling factor they should render at for best results.
  111. Link Mauve To avoid having to modify obscure config files.
  112. Holger I also see how this allows for a single knob to scale everything globally, which is nice, but not really my problem either.
  113. Link Mauve Here on Weston, every client renders correctly, no matter which toolkit it uses.
  114. Holger I.e. so far I haven't seen an excuse for GTK/Qt failing to scale properly on X11 :-)
  115. Link Mauve It just looks better when using GTK+ or GLFW, which properly support that feature.
  116. Link Mauve Qt5 still has many bugs, although there is now a gtkplatform QPA for it which already improves things.
  117. Link Mauve Holger, mostly just that X11 doesn’t expose that information.
  118. lovetox Link Mauve, do you know is a scale higher than 2 already common?
  119. Link Mauve Hmm, maybe on some mobile phones?
  120. Link Mauve I’ve never encountered it myself.
  121. Link Mauve I stay pretty far away from anything phone, though.
  122. lovetox of course, anybody using gajim on a phone is a maniac
  123. Link Mauve I’ve done that in the past. :x
  124. Link Mauve Aside from the very annoying popups, it was working quite well.
  125. ThibG That was even more crazy back then
  126. Link Mauve Was it?
  127. ThibG I mean, scaling the UI definitely wasn't the issue ;)
  128. Link Mauve Right.
  129. Link Mauve Although at 480×640 2.8" I had to use an adapted GTK+ theme.
  130. Link Mauve The Hackable:Devices one was pretty nice.
  131. bot Philipp Hörist pushed 1 commit to branch _refs/heads/master_ of _gajim_ <https://dev.gajim.org/gajim/gajim>: *2c7861bc* <https://dev.gajim.org/gajim/gajim/commit/2c7861bc1de8f8b0547e02b4ca8fd167c1018f41> Groupchat: Dont update avatar on offline presence
  132. Asterix prosody upgrade ...
  133. mdosch In the "server info" it would be nice if clicking to it would open the related xep in the browser :)
  134. lovetox good idea mdosch :D
  135. Asterix about #8768: We have for since .... The begining of Gajim. Nobody complains, why would we remove that? It's quite common to have *HEY* rendered as bold
  136. mdosch _HEY_?
  137. Asterix and finally /hey/
  138. mdosch Hehe, in my early internet days I was familiar to *bold* but now as I am markdown infected this is *italic* and this is **bold**
  139. mdosch we need a standard for this
  140. mdosch https://xkcd.com/927/
  141. bot pushed commits to branch __ of _gajim_ <https://dev.gajim.org/gajim/gajim>:
  142. Asterix it was a test
  143. bot pushed commits to branch __ of _gajim_ <https://dev.gajim.org/gajim/gajim>:
  144. Asterix hmm same msg when we test other events :/
  145. bot pushed commits to branch __ of _gajim_ <https://dev.gajim.org/gajim/gajim>:
  146. bot pushed commits to branch __ of _gajim_ <https://dev.gajim.org/gajim/gajim>:
  147. doczkal mdosch, suprisingly you are still familiar with plain/text writing... Unfortunately E-Mailing has moved so far away from plain/text that it's getting hard to actually read some mails :/
  148. mdosch Are you talking about html emails?
  149. doczkal mdosch, yes, or whatever Outlook scrambles into it's mails
  150. doczkal mdosch, Outlook does "some sort of html" but it's far away from HTML already
  151. bot pushed commits to branch __ of _gajim_ <https://dev.gajim.org/gajim/gajim>:
  152. bot pushed commits to branch __ of _gajim_ <https://dev.gajim.org/gajim/gajim>:
  153. debacle I have a small Python script, that sends email in both formats: A short, colourful HTML, saying "unfortunately, your email client does not support text email" and the second part with the email text. Order is important. The standard says, the last part (here: text) has precedence.
  154. bot pushed commits to branch __ of _gajim_ <https://dev.gajim.org/gajim/gajim>:
  155. doczkal debacle: hmm, usually it's the other way around if they bother at all and not just send html only emails