-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Vincent Ladeuil wrote: >>>>>> "vila" == Vincent Ladeuil <v.ladeuil...@free.fr> writes: > > > Jelmer> We rely on this in a lot of places, and we can't get > Jelmer> rid of it in the near future, and perhaps > Jelmer> never. Progress bars can be created by pretty much > Jelmer> any call to bzrlib. > > vila> I will tend to answer to that by: > > vila> if we have a progress widget: > vila> use it > vila> else: > vila> tell bzr to *not* report progress > > vila> And try to add progress widgets anywhere it's > vila> needed. This can be done progressively driven by demand > vila> (bzr doesn't report progress everywhere it should > vila> either). I don't think it's bad to have such progress windows in some cases. Opening a Branch/Repository can take ages in some situations and I think it's ok to just show a progress window in that case before showing the window that is opening that branch.
Some progress indication is better than none at all imho, it just makes the tool seem irresponsive. > Jelmer> If you for example run "bzr gannotate" in a subversion > Jelmer> working copy you get a window-based progress bar. > > Gannotate should defines its own progress widget. Yes. :-) > But I don't have a subversion working copy handy, do you have > another example *I* can use to reproduce and test the > window-based progress bar ? I think you may also be able to reproduce it e.g. when annotating files in the openoffice.org repository. > vila> But describing ad-hoc tests rocks ! How did you test ? :-) > > Still standing, I *think* my patch is correct in that respect, > but I still feel that the overall design (past *and* present) is > not bullet-proof. > > Jelmer> There's some bits of the network activity still being > Jelmer> printed to stdout. > > That's very strange because the default ui don't do that, only > TextUIFactory can produce that so far, and bzr-gtk installed its > own ui via open_display which should be called very early for any > command. > > Again, a recipe to reproduce ? I can't reproduce it with your latest patch anymore, but I'm sure I got a single line of output with activity reports earlier. _progress_all_finished shouldn't reset the progress bar widget to None; if the window triggers more progress reports, we shouldn't be displaying those in a window. To reproduce, try hitting "Refresh" in the viz window. Other than this, this patch seems good enough to land. We can always fix the minor issues later, right now this not landing breaks bzr-gtk in general. bb:tweak Cheers, Jelmer -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.9 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iJsEAQECAAYFAkoyab8ACgkQDLQl4QYPZuVROAP42cCR04Aa9e5qT+HgQ5nay+Om eOIESouEnY2smKBJ7z7STlUexdf7nPcvtTGG3iH2nFd0tCTDQ25/dkjRygSCTXxQ yvyFsGFZk9hr/Q+8mpiWRCB78V9WU2E3k0Bj2PLMcUv9l6duDdNgM1x/yiMXrEFy PyKVfb2UCYS+pSVGKQ== =0bxW -----END PGP SIGNATURE----- -- bzr-gtk mailing list bzr-gtk@lists.ubuntu.com Modify settings or unsubscribe at: https://lists.canonical.com/mailman/listinfo/bzr-gtk
