Some more bugs :
- There also is a weird alignment is the Transition pane (see screenshot attached)
- There is no cross but a pixel at the top right corner of the player to allow the user to dock it (see screenshot attached)
- With the wizard, when clicking on "Browse Projects" button, a new window appears but then shift (see video attached)
- French translation is not complete (see the Title pane for an exemple)

Created attachment 273274 [details]
weird alignment 4 Transition and Title

Created attachment 273275 [details]
No cross to dock the player but a lonely pixel

Created attachment 273276 [details]
Browse projects pane is shifting

Conclusion on my debian system : from a user point of view, after only 2 minutes toying with Pitivi "beta" on my Debian, it looks far more buggy that 0.91 & 0.92 that were however called "alphas" !

And also :
- Media library shows buttons during a one second when Pitivi is launched (see screenshot attached)

Created attachment 273279 [details]
Media library shows buttons during a one second when Pitivi is launched

Attachment #273267 [details]:
- The transparent timeline is a bug in clutter-gtk somewhere, not caused by the 0.93 release per se
- Text wrapping in the infobars: first, you're not giving it enough space, second, the same issue affects 0.92. You just don't see it as much because the French translation had wider words that prevented extreme wrapping. 

Attachment #273274 [details] "weird alignment in the Transition pane":
- That one might be an actual bug yes, should be filed separately :)

Attachment #273275 [details] "No cross to dock the player but a lonely pixel":
- 0.92 is exactly the same. That's how GTK/Adwaita represents close buttons on utility windows. It's a theme issue, not Pitivi.

Attachment #273276 [details] "open project file chooser shifts around":
- 0.92 is the same, just do Ctrl+O and you'll see. Not actually a regression introduced by 0.93. And probably not even a bug in Pitivi, I just looked at the "openProject" method in mainwindow.py and we set the transient_for property ("attach to the main window") right at the beginning, and call "run" on the dialog at the end of the openProject method, so there's nothing more we can do here.

Attachment #273279 [details] "Media library shows buttons during one second on startup":
- Not reproducible here. I've only seen that happen yesterday in Alexandru's version of the headerbar branch, which makes me think this is another race condition.


> Conclusion on my debian system : from a user point of view, after only 2
> minutes toying with Pitivi "beta" on my Debian, it looks far more buggy that
> 0.91 & 0.92 that were however called "alphas" !

As demonstrated, none of these actually are regressions introduced by 0.93.

You should not overlook the quality/stabilization work done in a release by being fooled by superficial things that are not even Pitivi's fault (they look like race conditions in GTK). All of the issues you have shown are cosmetic and have nothing to do with being buggier than previous releases.

(In reply to comment #8)
> Attachment #273267 [details]:
> - The transparent timeline is a bug in clutter-gtk somewhere, not caused by the
> 0.93 release per se

So, what is your advice ? Shall I fill a bug against libclutter-gtk-1.0-0 on Debian ?

(In reply to comment #8)
> Attachment #273274 [details] "weird alignment in the Transition pane":
> - That one might be an actual bug yes, should be filed separately :)

Ok, I'll do that !

> > Conclusion on my debian system : from a user point of view, after only 2
> > minutes toying with Pitivi "beta" on my Debian, it looks far more buggy that
> > 0.91 & 0.92 that were however called "alphas" !
> 
> As demonstrated, none of these actually are regressions introduced by 0.93.
> 
> You should not overlook the quality/stabilization work done in a release by
> being fooled by superficial things that are not even Pitivi's fault (they look
> like race conditions in GTK). All of the issues you have shown are cosmetic and
> have nothing to do with being buggier than previous releases.

My goal was not to be offensive, I'm sorry if you took it that way. 
I wanted to point out that these cosmetics bugs are the 1st experience that the user get with Pitivi (no real work was even done within Pitivi while reporting theses bugs) and that it could be bad for Pitivi reputation considering that you advertised a beta status whereas former versions were in alpha status. BUT I took good note that almost none of these bugs were new or pitivi bugs, therefore my point is null and void

adding some info gathered while testing 0.93-2:

Importing a clip from library to timeline, triggers python exception:

Traceback (most recent call last):

+ Trace 233413

Looking in gir1.2-cogl-1.0 (1.18-0-2) source wich is from cogl-1.18.0:

Source: cogl
Binary: libcogl20, libcogl-common, libcogl-dev, libcogl20-dbg, libcogl-doc, gir1.2-cogl-1.0, libcogl-path20, libcogl-path-dev, libcogl-pango20, libcogl-pango-dev, libcogl-pango20-dbg, gir1.2-coglpango-1.0, libcogl-gles2-20, libcogl-gles2-dev
Architecture: any all
Version: 1.18.0-2
Maintainer: Rico Tzschichholz <ricotz@ubuntu.com>
Uploaders: Andreas Henriksson <andreas@fatal.se>, Debian GNOME Maintainers <pkg-gnome-maintainers@lists.alioth.debian.org>, Emilio Pozuelo Monfort <pochu@debian.org>, Sjoerd Simons <sjoerd@debian.org>
Standards-Version: 3.9.5
Vcs-Browser: http://anonscm.debian.org/viewvc/pkg-gnome/desktop/unstable/cogl/
Vcs-Svn: svn://svn.debian.org/pkg-gnome/desktop/unstable/cogl

g -rn path_round_rectangle /opt/pitivi/deps/cogl-1.18.0/
> /opt/pitivi/deps/cogl-1.18.0/cogl-path/cogl-path.symbols:24:cogl_path_round_rectangle

thus symbol seems existing

then inspecting exposed methods from python:
>>> c = dir(Cogl)
>>> for elt in c:
...     print c
... 
'AFIRST_BIT'
'A_BIT'
'AttributeType'
'BGR_BIT'
'Bitmap'
'BitmapError'
'BlendStringError'
'BufferBit'
'BufferTarget'
'Color'
'ColorMask'
'DEPTH_BIT'
'DepthTestFunction'
'Euler'
'FIXED_0_5'
'FIXED_1'
'FIXED_2_PI'
'FIXED_BITS'
'FIXED_EPSILON'
'FIXED_MAX'
'FIXED_MIN'
'FIXED_PI'
'FIXED_PI_2'
'FIXED_PI_4'
'FIXED_Q'
'FeatureFlags'
'FilterReturn'
'Fixed'
'FogMode'
'IndicesType'
'Material'
'MaterialAlphaFunc'
'MaterialFilter'
'MaterialLayer'
'MaterialLayerType'
'MaterialWrapMode'
'Matrix'
'Offscreen'
'PREMULT_BIT'
'PixelFormat'
'Quaternion'
'RADIANS_TO_DEGREES'
'ReadPixelsFlags'
'RendererError'
'SQRTI_ARG_10_PERCENT'
'SQRTI_ARG_5_PERCENT'
'SQRTI_ARG_MAX'
'STENCIL_BIT'
'ShaderType'
'SystemError'
'TEXTURE_MAX_WASTE'
'Texture'
'TextureComponents'
'TextureError'
'TextureFlags'
'TextureType'
'TextureVertex'
'VerticesMode'
'Winding'
'WinsysFeature'
'_ColorSizeCheck'
'_MatrixSizeCheck'
'_TextureVertexSizeCheck'
'__class__'
'__delattr__'
'__dict__'
'__dir__'
'__doc__'
'__file__'
'__format__'
'__getattr__'
'__getattribute__'
'__hash__'
'__init__'
'__loader__'
'__module__'
'__name__'
'__new__'
'__path__'
'__reduce__'
'__reduce_ex__'
'__repr__'
'__setattr__'
'__sizeof__'
'__str__'
'__subclasshook__'
'__weakref__'
'_introspection_module'
'_load'
'_namespace'
'_overrides_module'
'_version'
'begin_gl'
'bitmap_error_quark'
'blend_string_error_quark'
'check_extension'
'clear'
'clip_ensure'
'clip_pop'
'clip_push'
'clip_push_rectangle'
'clip_push_window_rect'
'clip_push_window_rectangle'
'clip_stack_restore'
'clip_stack_save'
'clutter_check_extension_CLUTTER'
'clutter_winsys_has_feature_CLUTTER'
'color_equal'
'color_init_from_hsl'
'debug_matrix_print'
'disable_fog'
'double_to_int'
'double_to_uint'
'end_gl'
'features_available'
'flush'
'frustum'
'get_backface_culling_enabled'
'get_bitmasks'
'get_depth_test_enabled'
'get_features'
'get_modelview_matrix'
'get_projection_matrix'
'get_viewport'
'gtype_matrix_get_type'
'handle_get_type'
'handle_ref'
'handle_unref'
'is_bitmap'
'is_material'
'is_offscreen'
'is_program'
'is_shader'
'is_texture'
'is_vertex_buffer'
'is_vertex_buffer_indices'
'material_unref'
'matrix_equal'
'onscreen_clutter_backend_set_size_CLUTTER'
'ortho'
'perspective'
'polygon'
'pop_draw_buffer'
'pop_framebuffer'
'pop_matrix'
'pop_source'
'program_attach_shader'
'program_get_uniform_location'
'program_link'
'program_set_uniform_1f'
'program_set_uniform_1i'
'program_set_uniform_float'
'program_set_uniform_int'
'program_set_uniform_matrix'
'program_uniform_1f'
'program_uniform_1i'
'program_uniform_float'
'program_uniform_int'
'program_uniform_matrix'
'program_unref'
'program_use'
'push_draw_buffer'
'push_matrix'
'push_source'
'read_pixels'
'rectangle'
'rectangle_with_multitexture_coords'
'rectangle_with_texture_coords'
'rectangles'
'rectangles_with_texture_coords'
'rotate'
'scale'
'set_backface_culling_enabled'
'set_depth_test_enabled'
'set_draw_buffer'
'set_fog'
'set_modelview_matrix'
'set_projection_matrix'
'set_source'
'set_source_color'
'set_source_color4f'
'set_source_color4ub'
'set_source_texture'
'set_viewport'
'shader_compile'
'shader_get_info_log'
'shader_get_type'
'shader_is_compiled'
'shader_source'
'shader_unref'
'sqrti'
'texture_error_quark'
'transform'
'translate'
'vertex_buffer_add'
'vertex_buffer_delete'
'vertex_buffer_disable'
'vertex_buffer_draw'
'vertex_buffer_draw_elements'
'vertex_buffer_enable'
'vertex_buffer_get_n_vertices'
'vertex_buffer_indices_get_type'
'vertex_buffer_submit'
'vertex_buffer_unref'
'viewport'

...nothing about path_round_rectangle

Ok, now *that* is important information to actually determine that there's a bug in some distros... maybe some incorrect dependency. Retitling the bug report for clarity.

As someone mentioned in a comment, the Cogl 1.18 release removed the CoglPath API from the introspection file.
This was unintentional as a result of splitting the Cogl library into smaller pieces (cogl, cogl-path, cogl-pango, and the new cogl-gst) :

http://lists.freedesktop.org/archives/cogl/2014-April/001640.html

If someone could recompile the 1.18 branch of Cogl with the patches attached here :

http://lists.freedesktop.org/archives/cogl/2014-April/001641.html

And let me know if anything isn't working.

Sadly some modifications will be needed to the pitivi application to use the Path API again. Now most of the Path API should be in a CoglPath repository separate from the Cogl one. But apart from that it should behave the same.

Even though the incident in cogl was unintentional, we ended up pulling the dependency on rounded rectangles from Pitivi, as they were not actively used:


commit af65ef5317e2b7a0492ef165f2686c47b7011279
Author: Mathieu Duponchelle <mathieu.duponchelle@opencreed.com>
Date:   Wed Apr 2 20:56:08 2014 +0200

    elements: Don't use Cogl anymore.

commit c4c11dedd3cc966f8fa75fad080aff3abf1cf3f6
Author: Mathieu Duponchelle <mathieu.duponchelle@opencreed.com>
Date:   Wed Apr 2 18:09:14 2014 +0200

    elements: Don't use path_round_rectangle.
    
    We didn't actually make rounded rectangles anyway.
    
    This class will need to go altogether at some point.

I tried the daily build (http://fundraiser.pitivi.org/download-bundles) and the timeline is still transparent but only until the startup assistant is closed, then the timeline becomes solid

*** Bug 736345 has been marked as a duplicate of this bug. ***