https://bugzilla.redhat.com/show_bug.cgi?id=853590
Some options:
- nuke icedtea-web (this saves 100MB or so, but means people won't be able to test java in the live instance).
- seahorse - 7mb
- Longer term: split perl deps out of exo and see if we can remove perl? lm_sensors also needs perl, so we would have to drop xfce4-sensors-plugin, but I think it's worth it. That could save us a ton of room down the road. This dep is due exo-compose-mail-1 helper being a perl script.
- drop one of firefox or midori. ;)
- We really only need a bit of room at this point, so other smaller suggestions?
kevin
On Sat, Nov 10, 2012 at 10:39 PM, Kevin Fenzi kevin@scrye.com wrote:
- nuke icedtea-web (this saves 100MB or so, but means people won't be able to test java in the live instance).
+1 to remove java from the livecd, the size is to high for the limited web sites needing the java web plugin.
Tim
On Sat, 10 Nov 2012 14:39:37 -0700 Kevin Fenzi kevin@scrye.com wrote:
https://bugzilla.redhat.com/show_bug.cgi?id=853590
Some options:
- nuke icedtea-web (this saves 100MB or so, but means people won't
be able to test java in the live instance).
Have it somewhere on the spin page. You may need to install java..icedtea..
Am Samstag, den 10.11.2012, 14:39 -0700 schrieb Kevin Fenzi:
https://bugzilla.redhat.com/show_bug.cgi?id=853590
Some options:
- nuke icedtea-web (this saves 100MB or so, but means people won't be able to test java in the live instance).
It should be ~70 mb at least, but I am afraid it is unintuitive. People will just see that something doesn't work but hardly know how to fix it. "icedtea" is not really to find in the package management.
- seahorse - 7mb
Not a bad idea, however I consider seahorse useful.
- Longer term: split perl deps out of exo and see if we can remove perl? lm_sensors also needs perl, so we would have to drop xfce4-sensors-plugin, but I think it's worth it. That could save us a ton of room down the road. This dep is due exo-compose-mail-1 helper being a perl script.
I doubt that we cal really remove perl form the spin and even if we can, it means we loose some features like attaching a file form Thunar to a mail via the "Send to" menu.
- drop one of firefox or midori. ;)
Yes please! The question is: Which one? Throw a coin or play rock-paper-scissors(-spock)? ;)
- We really only need a bit of room at this point, so other smaller suggestions?
Drop some themes!
Kind regards, Christoph
On Sun, 11 Nov 2012 18:26:53 +0100 Christoph Wickert christoph.wickert@gmail.com wrote:
Am Samstag, den 10.11.2012, 14:39 -0700 schrieb Kevin Fenzi:
https://bugzilla.redhat.com/show_bug.cgi?id=853590
Some options:
- nuke icedtea-web (this saves 100MB or so, but means people won't be able to test java in the live instance).
It should be ~70 mb at least, but I am afraid it is unintuitive. People will just see that something doesn't work but hardly know how to fix it. "icedtea" is not really to find in the package management.
- seahorse - 7mb
Not a bad idea, however I consider seahorse useful.
- Longer term: split perl deps out of exo and see if we can remove perl? lm_sensors also needs perl, so we would have to drop xfce4-sensors-plugin, but I think it's worth it. That could save us a ton of room down the road. This dep is due exo-compose-mail-1 helper being a perl script.
I doubt that we cal really remove perl form the spin and even if we can, it means we loose some features like attaching a file form Thunar to a mail via the "Send to" menu.
- drop one of firefox or midori. ;)
Yes please! The question is: Which one? Throw a coin or play rock-paper-scissors(-spock)? ;)
- We really only need a bit of room at this point, so other smaller suggestions?
Drop some themes!
Kind regards, Christoph
I will second the "drop some themes"! It is an easy way to gain a bit of space.
If more than that is needed, flip the coin. Cutting one browser does not prevent anyone from installing it when wanted.
Am Sonntag, den 11.11.2012, 10:51 -0700 schrieb Charlie Kravetz:
I will second the "drop some themes"! It is an easy way to gain a bit of space.
If more than that is needed, flip the coin. Cutting one browser does not prevent anyone from installing it when wanted.
Right, and installing one's favorite browser is way more intuitive than installing icedtea-web.
Cheers, Christoph
On Sun, 11 Nov 2012 19:03:27 +0100 Christoph Wickert christoph.wickert@gmail.com wrote:
Am Sonntag, den 11.11.2012, 10:51 -0700 schrieb Charlie Kravetz:
I will second the "drop some themes"! It is an easy way to gain a bit of space.
If more than that is needed, flip the coin. Cutting one browser does not prevent anyone from installing it when wanted.
Right, and installing one's favorite browser is way more intuitive than installing icedtea-web.
Cheers, Christoph
I am sorry. I thought someone was asking for suggestions. I did not realize you already decided what to cut.
Am Sonntag, den 11.11.2012, 11:41 -0700 schrieb Charlie Kravetz:
On Sun, 11 Nov 2012 19:03:27 +0100 Christoph Wickert christoph.wickert@gmail.com wrote:
Am Sonntag, den 11.11.2012, 10:51 -0700 schrieb Charlie Kravetz:
Cutting one browser does not prevent anyone from installing it when wanted.
Right, and installing one's favorite browser is way more intuitive than installing icedtea-web.
I am sorry. I thought someone was asking for suggestions. I did not realize you already decided what to cut.
No need to be sorry, we didn't make a decision yet. It's not on me to decide anyways, it's on all of us. :)
I just wanted to point out that we both agree that installing a browser is easier than to figure out how to get Java working.
Kind regards, Christoph
On Mon, Nov 12, 2012 at 1:17 PM, Christoph Wickert < christoph.wickert@gmail.com> wrote:
Am Sonntag, den 11.11.2012, 11:41 -0700 schrieb Charlie Kravetz:
On Sun, 11 Nov 2012 19:03:27 +0100 Christoph Wickert christoph.wickert@gmail.com wrote:
Am Sonntag, den 11.11.2012, 10:51 -0700 schrieb Charlie Kravetz:
Cutting one browser does not prevent anyone from installing it when wanted.
Right, and installing one's favorite browser is way more intuitive than installing icedtea-web.
I am sorry. I thought someone was asking for suggestions. I did not realize you already decided what to cut.
No need to be sorry, we didn't make a decision yet. It's not on me to decide anyways, it's on all of us. :)
I just wanted to point out that we both agree that installing a browser is easier than to figure out how to get Java working.
Kind regards, Christoph
Hi all,
I think we should get rid of the browser which needs most space. If the removal of firefox does give us more space, then I would remove it or vice versa. If we stick to midori, we would probably also emphasize the fact of a lightweight spin. Also I think it's better to remove user-space programs than some packages like icedtea, where it's not so obvious what's missing. If firefox or midori is missing in the menu, it's kind of obvious to solve the "problem".
Just my opinion,
Johannes
xfce mailing list xfce@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/xfce
On Sun, 2012-11-11 at 18:26 +0100, Christoph Wickert wrote:
- drop one of firefox or midori. ;)
Yes please! The question is: Which one? Throw a coin or play rock-paper-scissors(-spock)? ;)
[x] drop Firefox [ ] drop Midori
On 11/12/2012 06:17 AM, Dominic Hopf wrote:
On Sun, 2012-11-11 at 18:26 +0100, Christoph Wickert wrote:
- drop one of firefox or midori. ;)
Yes please! The question is: Which one? Throw a coin or play rock-paper-scissors(-spock)? ;)
[x] drop Firefox [ ] drop Midori
Does midori work with the icedtea stuff? Currently, I'm asking myself, if we could drop one browser AND also icedtea?
Another good candidate to drop would be some fonts: - cjkuni-uming-fonts - wqy-zenhei-fonts both about 38 megs
What about opencv? (about 32 megs)
Or maybe we should discuss, if we drop printer drivers. In the Windows world, it seems so common to install a printer driver first, before you could configure it.
On Mon, 12 Nov 2012 12:39:15 +0100 Matthias Runge mrunge@matthias-runge.de wrote:
On 11/12/2012 06:17 AM, Dominic Hopf wrote:
On Sun, 2012-11-11 at 18:26 +0100, Christoph Wickert wrote:
- drop one of firefox or midori. ;)
Yes please! The question is: Which one? Throw a coin or play rock-paper-scissors(-spock)? ;)
[x] drop Firefox [ ] drop Midori
Does midori work with the icedtea stuff? Currently, I'm asking myself, if we could drop one browser AND also icedtea?
Yes, it does work with icedtea just fine.
Another good candidate to drop would be some fonts:
- cjkuni-uming-fonts
- wqy-zenhei-fonts
both about 38 megs
yeah, except then we drop looking nice in several locales. ;(
What about opencv? (about 32 megs)
Huh. Wonder what pulls that in... I'm happy removing it, and can't see what needs it off hand. Wait... I don't see it in the current spin. ;(
Or maybe we should discuss, if we drop printer drivers. In the Windows world, it seems so common to install a printer driver first, before you could configure it.
If there was an easy way to install them for users, sure... but that needs PackageKit, which we dropped in favor of yumex.
kevin
On 11/12/2012 09:58 PM, Kevin Fenzi wrote:
Huh. Wonder what pulls that in... I'm happy removing it, and can't see what needs it off hand. Wait... I don't see it in the current spin. ;(
Oh I took that https://bugzilla.redhat.com/show_bug.cgi?id=853590#c16
(that bz, you linked in the origin posting)
Or maybe we should discuss, if we drop printer drivers. In the Windows world, it seems so common to install a printer driver first, before you could configure it.
If there was an easy way to install them for users, sure... but that needs PackageKit, which we dropped in favor of yumex.
That's sad. Maybe we want to revert this decision? But: would that be a step forward? I really can't say.
On Tue, 13 Nov 2012 12:27:13 +0100 Matthias Runge wrote:
If there was an easy way to install them for users, sure... but that needs PackageKit, which we dropped in favor of yumex.
That's sad. Maybe we want to revert this decision? But: would that be a step forward? I really can't say.
Ah, yes, I would also be for keeping the packagekit related stuff in. True, PK looks a bit strange in a xfce environment but the "install missing codecs", "install detected printer's driver" and "click to install" are rather useful features... I cannot say anything about the main interface though -- I never liked yumex (but it's a couple of years since I last tried it) and PK also never grew on me -- looks like I'm stuck on command line yum forever, so I'm not very fit to give opinions on gui package managers.
Martin
On 11/13/2012 09:49 AM, Martin Sourada wrote:
Ah, yes, I would also be for keeping the packagekit related stuff in. True, PK looks a bit strange in a xfce environment but the "install missing codecs", "install detected printer's driver" and "click to install" are rather useful features... I cannot say anything about the main interface though -- I never liked yumex (but it's a couple of years since I last tried it) and PK also never grew on me -- looks like I'm stuck on command line yum forever, so I'm not very fit to give opinions on gui package managers.
Martin
yumex has click-to-install, the codecs thing isn't needed I suppose as any codec one might need is outside fedora repos so the one-click thing is already busted there.
Rests the driver one.
On Tue, Nov 13, 2012 at 12:49 PM, Martin Sourada martin.sourada@gmail.comwrote:
Ah, yes, I would also be for keeping the packagekit related stuff in. True, PK looks a bit strange in a xfce environment but the "install missing codecs", "install detected printer's driver" and "click to install" are rather useful features... I cannot say anything about the main interface though -- I never liked yumex (but it's a couple of years since I last tried it) and PK also never grew on me -- looks like I'm stuck on command line yum forever, so I'm not very fit to give opinions on gui package managers.
Martin
Martin, I am always open to suggestion to how to improve yumex, maybe your should try the current version, and let me know what you don't and how to improve it.
click to install has recently been implemented in yumex. install missing codecs is not very useful, because most of it exists outside fedora (rpmfusion), so most people has to go to rpmfusion.org to install the repos, so they also enter the one line yum command to install all the need gstreamer codecs. It is not something you do every day.
next mayor released will be using a yum based dbus daemon, there will make it much simpler to use for 3. party application to performing packaging action and is designed with simplicity in mind and not over engineered as PackageKit, there is trying to support every package engine there exists.
Tim
Hi Tim,
Martin, I am always open to suggestion to how to improve yumex, maybe your should try the current version, and let me know what you don't and how to improve it.
Sorry if I sounded too critical -- after all it's really been a long time since I tried yumex, and, after all, I seem to be more of a CLI guy when it comes to package management -- yumex didn't impress me (in the deep past), packagekit kept me interested for a while thanks to fancy updates, but I always fell back to just using yum... It seems that any hint of a UI either graphical or text (I recently tried some TUI on OpenSUSE and didn't like it one bit, same with debian about a year back) gets in my way no matter how complex, or simple it is.
Anyway, in the case I'd give it a spin then if I'd have some constructive critic (suggestions) by then, I'd definitely share it with you :)
click to install has recently been implemented in yumex.
That's good to hear as that's the only case where I'm rather grateful that GUI package manager exists.
install missing codecs is not very useful, because most of it exists outside fedora (rpmfusion), so most people has to go to rpmfusion.org to install the repos, so they also enter the one line yum command to install all the need gstreamer codecs. It is not something you do every day.
I hate to remember names of packages so I always used to do it in a way that I installed rpmfusion and then let the music/video players install the missing codecs on demand via PK. Though I'm not sure how much of them is actually able to do that -- I used preupgrade the last time and before then I was using rythmbox+totem+mplayer instead of quodlibet+parole+mplayer2...
Anyway, the bottom line is, if the mentioned "killer" features of PK do not work on XFCE spin (sans rpmfusion installation) then having it there gives no added value and I don't have any issues with having yumex instead.
Regards, Martin
...
next mayor released will be using a yum based dbus daemon, there will make it much simpler to use for 3. party application to performing packaging action and is designed with simplicity in mind and not over engineered as PackageKit, there is trying to support every package engine there exists.
Tim
I've added a couple of suggestions on github that I had noticed here.
On a side-note, I'm using the test package for 3.0.10 and I noticed that its 'working window' is a bit more verbose than in 3.0.9 with some messages like
Unknown metadata type (group_gz) .. Unknown metadata type (pkgtags) .. Unknown metadata type (other_db) .. Repo Metadata for XXXX.xml
appearing on occasions.
On Tue, 13 Nov 2012 12:49:26 +0100 Martin Sourada martin.sourada@gmail.com wrote:
On Tue, 13 Nov 2012 12:27:13 +0100 Matthias Runge wrote:
If there was an easy way to install them for users, sure... but that needs PackageKit, which we dropped in favor of yumex.
That's sad. Maybe we want to revert this decision? But: would that be a step forward? I really can't say.
Ah, yes, I would also be for keeping the packagekit related stuff in. True, PK looks a bit strange in a xfce environment but the "install missing codecs", "install detected printer's driver" and "click to install" are rather useful features... I cannot say anything about the main interface though -- I never liked yumex (but it's a couple of years since I last tried it) and PK also never grew on me -- looks like I'm stuck on command line yum forever, so I'm not very fit to give opinions on gui package managers.
I'm not sure those features actually work under Xfce in any case.
I know there's no updates indicator in Xfce. Also, I think PackageKit pulls in a bunch more gnome stuff we don't want/use.
I also use yum (or dnf if I am feeling dangerous), but I think yumex is a better fit for our spin.
kevin
On Tue, 13 Nov 2012 10:09:16 -0700 Kevin Fenzi wrote:
I'm not sure those features actually work under Xfce in any case.
I know there's no updates indicator in Xfce. Also, I think PackageKit pulls in a bunch more gnome stuff we don't want/use.
Ah, that's bad... I wish gnome people would make their apps more usable outside the gnome environment :-(
I also use yum (or dnf if I am feeling dangerous), but I think yumex is a better fit for our spin.
Well, if the features does not work, than it's a no-brainer... But I vaguely recall that at least the printer driver install worked for me.
Martin
On 11/13/2012 03:09 PM, Kevin Fenzi wrote:
I know there's no updates indicator in Xfce.
If there's one for Mate (and other conveniences) it can be installed with https://github.com/linuxmint/xfce4-xfapplet-plugin Might be worth packaging.↑
On Tue, 13 Nov 2012 20:42:37 -0200 Sergio secipolla@gmail.com wrote:
On 11/13/2012 03:09 PM, Kevin Fenzi wrote:
I know there's no updates indicator in Xfce.
If there's one for Mate (and other conveniences) it can be installed with https://github.com/linuxmint/xfce4-xfapplet-plugin Might be worth packaging.↑
It was, we dropped it when gnome2 was dropped, I suppose we could bring it back if mate folks package up the applets again.
Are you sure update indicators work in mate? My understanding is that PackageKit no longer has any systray support, it uses gnome3 notifications only, so I would expect it to not work in mate either.
kevin
On 11/13/2012 08:56 PM, Kevin Fenzi wrote:
On Tue, 13 Nov 2012 20:42:37 -0200 Sergio secipolla@gmail.com wrote:
On 11/13/2012 03:09 PM, Kevin Fenzi wrote:
I know there's no updates indicator in Xfce.
If there's one for Mate (and other conveniences) it can be installed with https://github.com/linuxmint/xfce4-xfapplet-plugin Might be worth packaging.↑
It was, we dropped it when gnome2 was dropped, I suppose we could bring it back if mate folks package up the applets again.
Are you sure update indicators work in mate? My understanding is that PackageKit no longer has any systray support, it uses gnome3 notifications only, so I would expect it to not work in mate either.
I don't know. Isn't there a mate-packagekit ;-)
So, heres the current top 40 packages by size:
122862271 kernel 117821428 glibc-common 93833830 java-1.7.0-openjdk 40656267 webkitgtk 40381410 xulrunner 35623360 linux-firmware 35475387 mesa-dri-drivers 34904384 perl 33147198 foomatic-db-ppds 29809732 firefox 29454189 gnumeric 28363236 libpinyin-data 27879319 libpurple 26038006 anthy 25009477 python-libs 22247740 libicu 21457344 llvm-libs 21094325 cjkuni-uming-fonts 20990480 gutenprint 20543461 libabiword 18509337 grub2-efi 17156348 wqy-zenhei-fonts 17036975 ghostscript 16501168 samba-libs 16088019 selinux-policy-targeted 15989135 binutils 15914825 grub2-tools 15147474 claws-mail 14233669 coreutils 14149244 iso-codes 13845828 glibc 13351916 gtk2 12803865 nhn-nanum-gothic-fonts 12632645 geany 12370754 gtk3 12012878 poppler-data 11713752 sssd 10076667 spherical-cow-backgrounds-single 9739522 glib2 9638099 gnome-icon-theme
Looking at this... what about removing gnumeric? Most folks that need spreadsheets aren't going to use a live media, even then they would probibly want to test the libreoffice version? gnumeric has no deps and is 28MB.
Any objections?
kevin
On Thu, Nov 15, 2012 at 11:20 PM, Kevin Fenzi kevin@scrye.com wrote:
Looking at this... what about removing gnumeric? Most folks that need spreadsheets aren't going to use a live media, even then they would probibly want to test the libreoffice version? gnumeric has no deps and is 28MB.
Any objections?
+1, On of the first things i remove after installing the Xfce Live cd.
Tim
This won't have any effect on space but while looking on gstreamer stuff I found that gstreamer-plugins-espeak is installed on the live-cd but nothing requires it (it's for sugar).
On Sun, 11 Nov 2012 18:26:53 +0100 Christoph Wickert christoph.wickert@gmail.com wrote:
Am Samstag, den 10.11.2012, 14:39 -0700 schrieb Kevin Fenzi:
https://bugzilla.redhat.com/show_bug.cgi?id=853590
Some options:
- nuke icedtea-web (this saves 100MB or so, but means people won't
be able to test java in the live instance).
It should be ~70 mb at least, but I am afraid it is unintuitive. People will just see that something doesn't work but hardly know how to fix it. "icedtea" is not really to find in the package management.
indeed. ;(
- seahorse - 7mb
Not a bad idea, however I consider seahorse useful.
I don't use it here, so I would be fine with it going away...
- Longer term: split perl deps out of exo and see if we can remove perl? lm_sensors also needs perl, so we would have to drop xfce4-sensors-plugin, but I think it's worth it. That could save
us a ton of room down the road. This dep is due exo-compose-mail-1 helper being a perl script.
I doubt that we cal really remove perl form the spin and even if we can, it means we loose some features like attaching a file form Thunar to a mail via the "Send to" menu.
Well, we would need to re-implement that... Probibly in python if not c. ;(
- drop one of firefox or midori. ;)
Well, I use midori, so you know my vote. ;)
Happy to go the other way if people like tho.
Drop some themes!
We could, I guess that would be fine too.
kevin
Kevin Fenzi kevin at scrye.com Sat Nov 10 21:39:37 UTC 2012
Bug is closed, so I respond here.
The guys of aptosid managed to do a Xfce media image that has not at all the size of a full CD. KDE full needs >2GB but with an older version (4.8.4 - F18 is known to ship the 4.9 version).
Maybe we can steal some good ideas to shrink the image? ;)
distrowatch.eu: aptosid-2012-01-kde-full-i386-amd64.iso" (2,054MB) aptosid-2012-01-xfce-i386.iso (509MB) aptosid-2012-01-xfce-amd64.iso (515MB)
BG, Raphael
Not a problem to reduce size, the problem is to find that functionality to cut away :)
On Sun, Dec 2, 2012 at 2:57 PM, Raphael Groner raphgro@web.de wrote:
Kevin Fenzi kevin at scrye.com Sat Nov 10 21:39:37 UTC 2012
Bug is closed, so I respond here.
The guys of aptosid managed to do a Xfce media image that has not at all the size of a full CD. KDE full needs >2GB but with an older version (4.8.4 - F18 is known to ship the 4.9 version).
Maybe we can steal some good ideas to shrink the image? ;)
distrowatch.eu: aptosid-2012-01-kde-full-i386-amd64.iso" (2,054MB) aptosid-2012-01-xfce-i386.iso (509MB) aptosid-2012-01-xfce-amd64.iso (515MB)
BG, Raphael _______________________________________________ xfce mailing list xfce@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/xfce
On 12/02/2012 12:18 PM, tim.lauridsen@gmail.com wrote:
Not a problem to reduce size, the problem is to find that functionality to cut away :)
On Sun, Dec 2, 2012 at 2:57 PM, Raphael Groner raphgro@web.de wrote:
Kevin Fenzi kevin at scrye.com Sat Nov 10 21:39:37 UTC 2012
Bug is closed, so I respond here.
The guys of aptosid managed to do a Xfce media image that has not at all the size of a full CD. KDE full needs >2GB but with an older version (4.8.4 - F18 is known to ship the 4.9 version).
Maybe we can steal some good ideas to shrink the image? ;)
distrowatch.eu: aptosid-2012-01-kde-full-i386-amd64.iso" (2,054MB) aptosid-2012-01-xfce-i386.iso (509MB) aptosid-2012-01-xfce-amd64.iso (515MB)
BG, Raphael
I didn't check the aptosid isos but Debian splits a lot the sources so that tends to make it easier to do a smaller installation.