[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: LAST REMINDER: Debian Installer team meeting TOMORROW

Davide Viti wrote:
On Fri, 16 Sep 2005 14:40:37 +0200, Colin Watson wrote:

On Fri, Sep 16, 2005 at 07:13:59AM +0200, Christian Perrier wrote:

-status of the graphical installer
			   need Attilio, Colin at the minimum

I really haven't been paying much attention to the gtk frontend lately.

As for me, it's almost the same: after a couple of months of struggle
during the summer, I've been rather busy at work during the last few weeks
and I couldn't follow the developement very closely.

Most of my time was spent following the developement of gtk/directfb:
gtk+ 2.8 has finally been released (actually 2.8.3) and it happened at the
same time as Mike Emmel went on holiday (he's in charge of developing
gtk/direcfb). Haven't heard anything from him since the middle of August:
I'll ping him next week. Anyway: even if the status directfb seems good
enough for the installer, looks like there's a lot of work to do in order
to bring all the related udebs up to date. In the meanwhile Eugeniy gave
an easy and ready solution to simplify testing and developement of the
graphical fe based on a minimal X server: it turned out to be very useful
but it's still not clear what will be used... X or directfb. I'm not sure
I'll be attending tomorrow, since weekends for me are usually away from my
computer; hope to be there

Hi everybody

before anything else i'm pleased to tell you that yesterday i finally graduated: my thesis on the graphical d-i was very appreciated so i would like to thank the many of you who helped me so much in this work (cjw, bubule, fjp, zino, joey and the others i'm forgetting right now :). I hope i'll be able to attend to the meeting this afternoon, but since i'm having dinner out with my relatives i cannot tell for sure i'll be there. In case i won't be able to attend at the meeting here are some toughts on the graphical d-i.

-The frontend itself seems to work fine: no runtime warnings in X nor DFB, no strange crashes (apart from those related to the main-menu hack that can be disabled for compability sake. ) Work needs still to be done to prettify the interface and to make it comply with GNOME guidelines on how to write graphical interfaces (e.g: all those frames around questions should be removed) More work needs to be done to de-harcode some parameters that are now built-in inside the frontend: cdebconf.conf seems the right place to put GTK frontend related parameters that need to be parsed at runtime.
Some (normal/minor) bugs are still awaiting to be closed.
Even if my thesis is now over i'd like to keep on working on the GTK frontend until it's finished.

On the contraty hard work needs to be done on many other fields:

-X or DFB ? ATM the GTK frontend works fine in conjunction with GDKDFB 2.0.9-5 , but the lack of a libgtk+2.0-directfb0-udeb udeb is preventing us from creating a graphcal netinst cd. Alastair should be working on this right now, but i haven't had news from him since last week. On the DFB side we'll be stuck with GTK 2.0.9 for a long time: as Davide said Mike Emmel is still on vacation, so we'll have to wait some weeks/months before a new GDK DFB version is ready to compile with GTK 2.8.x . GTKDFB 2.0 is enough, at the moment, for the graphical d-i needs, but for sure it would be better lay on GTK 2.8.x as soon as possible. As soon as GTKDFB will be ready some more udebs will need to be prepared: libcairo, libice, ... Since we need an "embedded-like" approach for the GTK frontend IMHO i think DFB is better that X in this particular case because of its overall lower memory requirements. Still i have to admit that X is more stable, mature and functional than DFB: we may need to decide to switch to X one day if the GDK DFB developement had to stop because of any reasons (at the moment there is just a worst-case scenario, since Mike Emmel dimonstrated interest in making GDKDFB work on GTK 2.8.x as soon as possible ). In this case David Nusinow should be able to prepare an appropriate XFree udeb, but someone will be needed to prepare GTK, CAIRO, .. udebs. Anyway, if we want to display some nice logos/banners/?? in the GTK frontend someone should draw them (and i'm not an artist at all).

-FONTS: this is crucial point for the "Debian world domination project".
At themoment we only have ttf-freefont-udeb and, as you may have read in recent posts on debian-boot, debate is still open on how to pack all the needed fonts. What's better? an omni-comphrensive font package with all the neded glyphs or many smaller udebs where each one contains a part of the needed unicode glyphs? In the latter case we'll need to prepare a fontconfig configuration file to use proper glyphs from eack font wherever they may overlap; it would be also useful to be able to unload unneded font udebs when switching between languages (some font packages like ttf-arphic-uming (12/18 megs.) could be real pain).

-We'll need also to decide what's the minimum amount of memory needed to make the GTK frontend run: 128 megs. could be a reasonable value and at boot time an appropriate script should decide what fronend (NEWT/GTK) has to be started.

I don't know if this always hapens, but i've seen that when things appear to start to working other people more glady tends to help out in completeng what has still to be done. So i suggest we should create first some demo install cdroms as soon all basic components needed to make it work are ready, even if we only have ugly TT fonts and ugly GTK frontend, and "publicize" it. This way more users/developers should get interest in the project and they will eventually help out speeding up the development process.



Reply to: