Flawed approach in crunchbang-desktop & co.

EmaRsk · 2009-01-22 01:11:49

Let's start with an important statement: I like CrunchBang Linux, and my criticism wants to be absolutely constructive.
That made clear, I must confess that I'm glad I didn't install the crunchbang-desktop package without first downloading and inspecting it, because what I saw was very unpleasant for me.
That package (and its siblings crunchbang-desktop-lite and cruncheee-desktop, crunchbang-datafiles and crunchbang-lite-datafiles) installs a tarballed set of files (/usr/share/CrunchBang/CrunchBang.tar.gz) and then, with a postinstall script, it untars the files and copies them into the assigned places (/etc/, /root/, /usr/share/) without asking anything, and, even worse, with no possibility to undo things when uninstalling, because (remember?) the only regularly installed file is the tarball (plus the usual /usr/share/doc/ files).
This is the wrong approach! Period.
The right approach would be to let dpkg install every file (what do you think a package manager is for?), using diversions where necessary, and use a debconf dialog to tell the user to copy the new /etc/skel/ files into the home directory.
I'm sorry I don't have enough time right now to submit a corrected .deb package, but I'll try to do that as soon as possible.

flicck · 2009-01-22 01:52:25

Yeah, that sure seems like a whole lot of NOT setting things up for APT and dpkg to deal meaningfully with the files those packages end up dropping onto the system. I don't know much about Debian packaging, but I know the whole point of Debian's package management is that it is supposed to manage the installation and removal of packages' files not just packages.

Important distinctions to learn about : is this a case of the author(s) of these packages knowing about proper Debian packaging but choosing to not do so ( for whatever reason ), or a case of the author(s) not knowing? Either truth has implications/consequences.

corenominal · 2009-01-22 09:58:41

Hello EmaRsk smile

I appreciate your concerns and fully understand why you have raised this as an issue. The state of the current desktop packages are not ideal, but every effort is being made to modify the packages so that they comply with the Debian policy for packaging. For what it is worth, the packages are a hang-up from the first release, which was never really intended to be used by anyone other than myself. I can honestly say, I never thought for one moment that the release would grow into what it has become. I will be working on the packages again this weekend, hopefully I will remove all legacy hacks and improve the packages so that they comply with the Debian way of doing things.

As for:

EmaRsk wrote:

This is the wrong approach! Period.

Again, I fully understand and agree with what you have said and I will make every effort to improve the current packages, but consider this fact, while CrunchBang relies on both Debian and Ubuntu, it is actually neither. Thank you again for your constructive feedback, it really is appreciated! smile

flicck wrote:

Important distinctions to learn about : is this a case of the author(s) of these packages knowing about proper Debian packaging but choosing to not do so ( for whatever reason ), or a case of the author(s) not knowing? Either truth has implications/consequences.

flicck, this is a case of the author being aware of how things should be done, but for whatever reason, choosing to do it a different way. Regarding the implications/consequences, I would like to inform you that there is nothing to worry about, but that would be remiss of me; it is for you to decide, the files in question are all available for inspection and you are welcome to satisfy any curiosities you may have. smile

Final note (although I am sure this thread has the potential to run): The files being discussed are mainly customisations to existing configuration files created/installed by packages coming from the Ubuntu repositories. The current desktop-packages contain a tarball which is extracted to /usr/share/CrunchBang. Files extracted to this directory are then copied to their respective homes by way of a post installation script. As pointed out by EmaRsk, this is not standard practice. Since the original release, the desktop packages have been modified/changed on numerous occasions and reduced in size as files have been moved out of the desktop packages and into packages of their own. The process is ongoing and eventually the desktop packages should in fact contain/install a single data file, "/etc/crunchbang-lsb-release".

If anyone has any concerns about what is being discussed here, please do feel free to post any questions. smile

supersponge · 2009-01-22 10:07:29

I won't pretend to fully understand exactly what this entails, but it does interest me. Are the proposed changes going to be in place by the next release (whenever that may be)?

I know this is off topic, but what kind of release cycle are you looking for? Are you going to mimic the ubuntu release cycle, following closely behind the official releases?

corenominal · 2009-01-22 10:17:25

supersponge wrote:

I won't pretend to fully understand exactly what this entails, but it does interest me. Are the proposed changes going to be in place by the next release (whenever that may be)?

Yes. smile

I know this is off topic, but what kind of release cycle are you looking for? Are you going to mimic the ubuntu release cycle, following closely behind the official releases?

There is currently no release schedule in place. So far I have tried to follow the Ubuntu releases, but due to the limited resources/time I am not able to guarantee when a release comes out. If I do not make any promises, I cannot break them. wink Seriously though, a new project has been created on Launchpad, it is still early days but hopefully this type of query will be discussed and something more concrete put in place. If more people come aboard to help out with package creation and what not, a solid release schedule could become a reality. smile

Nik_Doof · 2009-01-22 10:27:00

Essentially to get full and proper coverage and correct methods of doing thinks, we're looking at forking each of the packages that need the config changes and creating a #! version. That is, if the exiting crunchbang-data collides with any existing packages, otherwise we can get away with a "crunchbang-openbox-data" type package.

rizzo · 2009-01-22 11:17:07

Nik_Doof wrote:

Essentially to get full and proper coverage and correct methods of doing thinks, we're looking at forking each of the packages that need the config changes and creating a #! version.

Hence we get back to the idea of a launchpad team and looking for packagers to help share the work load.

sewmyheadon · 2009-01-22 17:04:24

omns wrote:

Nik_Doof wrote:
Essentially to get full and proper coverage and correct methods of doing thinks, we're looking at forking each of the packages that need the config changes and creating a #! version.
Hence we get back to the idea of a launchpad team and looking for packagers to help share the work load.

I'd be happy to help with packaging, although I'll likely need a little tutoring or guidance to get started. smile

I'd definitely feel more like a worthwhile contributor if I could help by taking responsibility for a few #! packages.

michaelramm · 2009-01-22 17:29:03

I agree with sewmyheadon...I don't know anything about packaging but would be willing to learn.

I like the Arch approach...list all the packages and let folks pick the ones that they want to maintain. That way you get folks involved with packages that they actively want to see maintained and get better.

Michael

sewmyheadon · 2009-01-22 17:31:38

michaelramm wrote:

I like the Arch approach...list all the packages and let folks pick the ones that they want to maintain. That way you get folks involved with packages that they actively want to see maintained and get better.

Me too! I've worked with ZenWalk before and I believe they do/did something similar. Great way to get people working on things in which they have a vested interest.

michaelramm · 2009-01-22 17:41:43

sewmyheadon wrote:

michaelramm wrote:
I like the Arch approach...list all the packages and let folks pick the ones that they want to maintain. That way you get folks involved with packages that they actively want to see maintained and get better.
vested interest.

THAT was the phrase I was looking for...I spent 5 minutes trying to get that phrase, and never did.

Thanks,
Michael

flicck · 2009-01-24 00:57:30

@ corenominal : I had a suspicion that this might be a holdover from your original project, and the whole making CrunchBang for yourself deal. EmaRsk raised a valid concern/point, and I wanted to find out the whys and the wherefores. Thanks for the explanation!

flicck · 2009-01-25 16:15:44

corenominal : the reason I'm interested in this topic so much is pure self-interest : my "ideal dream linux distro" for years has been something that uses .debs/dpkg/APT, has newer software than Debian Stable but fewer showstopper bugs than Debian Unstable ( and even Testing shortly after a new Stable release is out ) and that allows for seamless upgrades between releases. Ubuntu has been very close to that for me, but never quite -- upgrading to new versions of the distro have always been problematic enough for me that fresh installs just worked better ( which was how I got interested in keeping separate /data partitions so my music, pictures, and such were always readily available ). Unless I'm mistaken, the issues being discussed in this thread are the reason there need to be "upgrade instructions" between releases of #!, as seems to be evidenced here : http://crunchbanglinux.org/forums/post/6795/#p6795

#! is ridiculously close to how I'd set up a desktop Linux system out of the box. The community is terrific. I'd love to watch this last tipping point for me, well, get tipped, so to speak. Again, total self-interest. So as to NOT just be a thoroughly selfish jerk, I see that you plan to address these issues with the packaging for the next release, and seeing as I've wanted to learn the nuts and bolts of Debian packaging but hadn't had any real impetus to learn ( everything I'm interested in running has already been neatly packaged up by someone ), I'd love to help/contribute to the resolution of these packaging issues. How best could I help you out?

corenominal · 2009-01-26 04:17:52

flicck wrote:

Unless I'm mistaken, the issues being discussed in this thread are the reason there need to be "upgrade instructions" between releases of #!, as seems to be evidenced here : http://crunchbanglinux.org/forums/post/6795/#p6795

Partly, the issues experienced in that post were due to a change in the build process between 01 and 02 -- this is actually fixable without too much effort, however I am hesitant to continue with working on the current packages without first fixing the problems identified in this thread. The upgrade process is without a doubt my biggest concern at the moment and it is something I would like to get sorted. So, it probably goes without saying, it is currently my top priority. wink

#! is ridiculously close to how I'd set up a desktop Linux system out of the box. The community is terrific. I'd love to watch this last tipping point for me, well, get tipped, so to speak. Again, total self-interest. So as to NOT just be a thoroughly selfish jerk, I see that you plan to address these issues with the packaging for the next release, and seeing as I've wanted to learn the nuts and bolts of Debian packaging but hadn't had any real impetus to learn ( everything I'm interested in running has already been neatly packaged up by someone ), I'd love to help/contribute to the resolution of these packaging issues. How best could I help you out?

Thank you for this, it means a lot and I would be grateful for any assistance. smile As for the best way to help out, if you have not done so already, please consider joining the CrunchBangers team on Launchpad. The team has a mailing list for developer discussions and it would be good to have you aboard. wink

kBang · 2009-01-26 05:10:10

corenominal wrote:

Partly, the issues experienced in that post were due to a change in the build process between 01 and 02 -- this is actually fixable without too much effort, however I am hesitant to continue with working on the current packages without first fixing the problems identified in this thread. The upgrade process is without a doubt my biggest concern at the moment and it is something I would like to get sorted. So, it probably goes without saying, it is currently my top priority.

With the current growing interest in this distribution (I recall at one point you didn't even like to refer to it as a distribution since it was your personal spin on Openbox+Ubuntu), I think your priorities are in good order, Philip.

I am just a tad concerned you may become overwhelmed with your success and become overworked on a project you originally didn't intend to grow so "large".

dannytatom · 2009-01-26 05:12:30

kBang wrote:

I am just a tad concerned you may become overwhelmed with your success and become overworked on a project you originally didn't intend to grow so "large".

I too have been concerned about this the last few days. ;o

corenominal · 2009-01-26 07:42:17

dannytatom wrote:

kBang wrote:
I am just a tad concerned you may become overwhelmed with your success and become overworked on a project you originally didn't intend to grow so "large".
I too have been concerned about this the last few days. ;o

Thank you both for your concern. smile I have to admit that I am finding it quite difficult to keep track lately; I enjoy these forums immensely and there was a time when I could at least read every topic, if not reply to most. This is now becoming an impossibility and I am finding I can longer offer the kind of personal support I once could when the project existed as a personal venture. While this makes me a little bit sad, I am also happy to see the community growing and I have noticed that members are more than willing to help each other out with their queries/problems, which is fantastic as it allows me time to work on all the existing bugs, flaws and new developments. smile

debaser · 2009-01-26 12:09:10

Thank you both for your concern. I have to admit that I am finding it quite difficult to keep track lately; I enjoy these forums immensely and there was a time when I could at least read every topic, if not reply to most. This is now becoming an impossibility and I am finding I can longer offer the kind of personal support I once could when the project existed as a personal venture. While this makes me a little bit sad, I am also happy to see the community growing and I have noticed that members are more than willing to help each other out with their queries/problems, which is fantastic as it allows me time to work on all the existing bugs, flaws and new developments.

Corenominal -

I find your personal touch and tenacity in replying/contributing to threads on the forum really great. However - coupled with the development, site management, launchpad, etc. I really hope burnout is not approaching for you. Would another forum Mod and general distro liaisons (install support, packaging manager, art, stamp licker smile , etc.) be appropriate at this point..to share the load?

rizzo · 2009-01-26 13:02:04

corenominal wrote:

I enjoy these forums immensely and there was a time when I could at least read every topic

Yes, I'd have to agree. It seems only a few week ago (actually it was) that I could easily check in on every thread and keep up with what was happening. Growth has been rapid lately but as you say support is being picked up admirably by the community smile

kestrel · 2009-01-26 14:10:15

corenominal wrote:

dannytatom wrote:
kBang wrote:
I am just a tad concerned you may become overwhelmed with your success and become overworked on a project you originally didn't intend to grow so "large".
I too have been concerned about this the last few days. ;o
Thank you both for your concern. I have to admit that I am finding it quite difficult to keep track lately; I enjoy these forums immensely and there was a time when I could at least read every topic, if not reply to most. This is now becoming an impossibility and I am finding I can longer offer the kind of personal support I once could when the project existed as a personal venture. While this makes me a little bit sad, I am also happy to see the community growing and I have noticed that members are more than willing to help each other out with their queries/problems, which is fantastic as it allows me time to work on all the existing bugs, flaws and new developments.

Philip, while we all can't work on the inner plumbing of #!, we can help out by answering questions here and on IRC. Go ahead and shift your focus to the important topic of the upgrade process. We'll cover for you. I too have held back on upgrading to 8.10.2 because current process is confused.

flicck · 2009-01-27 03:33:24

corenominal : for what it's worth, I doubt if #! would be under such scrutiny if it wasn't attracting interest far and wide. And it deserves the interest, you've made a distro that I think strikes the perfect balance between revelling in geekery and ease of use, which is a balancing act that takes technical and aesthetic chops that I certainly don't have. I'll look into joining the Crunchbangers after I've read up on the Ubuntu wiki about packaging -- Nik Doof rightly points out that it would be a good idea for people to join only if they're really going to get on with actually doing something tangible for the project.

Or, to put it more simply : thank you, sir, for making such a terrific distro and letting all of us tinker with your stroke of genius.

gorem · 2009-01-27 10:23:38

It fantastic to see the community response to this distribution rise up from being a support network for Phillip in his endeavours at producing his own personal type Linux distribution to now being what it is.

But as each new version ticks by it becomes imperative that more infrastructure is developed to support this growth. To my delight it looks like this is happening and happening at a cracking pace.

While I may not be able to contribute to Crunchbangers in any technical capacity I really respect the dedication shown in choosing to tackle these issues EARLY in the life of this great distribution. Resolving these issues NOW means a positive foundation is being constructed for CrunchbangLinux to be a very successful distribution in its own right.

Kudos to you Phil and your team, keep up the good work and take it easy on yourself. wink

CrunchBang Linux

SEARCH

#1 2009-01-22 01:11:49

Flawed approach in crunchbang-desktop & co.

#2 2009-01-22 01:52:25

Re: Flawed approach in crunchbang-desktop & co.

#3 2009-01-22 09:58:41

Re: Flawed approach in crunchbang-desktop & co.

#4 2009-01-22 10:07:29

Re: Flawed approach in crunchbang-desktop & co.

#5 2009-01-22 10:17:25

Re: Flawed approach in crunchbang-desktop & co.

#6 2009-01-22 10:27:00

Re: Flawed approach in crunchbang-desktop & co.

#7 2009-01-22 11:17:07

Re: Flawed approach in crunchbang-desktop & co.

#8 2009-01-22 17:04:24

Re: Flawed approach in crunchbang-desktop & co.

#9 2009-01-22 17:29:03

Re: Flawed approach in crunchbang-desktop & co.

#10 2009-01-22 17:31:38

Re: Flawed approach in crunchbang-desktop & co.

#11 2009-01-22 17:41:43

Re: Flawed approach in crunchbang-desktop & co.

#12 2009-01-24 00:57:30

Re: Flawed approach in crunchbang-desktop & co.

#13 2009-01-25 16:15:44

Re: Flawed approach in crunchbang-desktop & co.

#14 2009-01-26 04:17:52

Re: Flawed approach in crunchbang-desktop & co.

#15 2009-01-26 05:10:10

Re: Flawed approach in crunchbang-desktop & co.

#16 2009-01-26 05:12:30

Re: Flawed approach in crunchbang-desktop & co.

#17 2009-01-26 07:42:17

Re: Flawed approach in crunchbang-desktop & co.

#18 2009-01-26 12:09:10

Re: Flawed approach in crunchbang-desktop & co.

#19 2009-01-26 13:02:04

Re: Flawed approach in crunchbang-desktop & co.

#20 2009-01-26 14:10:15

Re: Flawed approach in crunchbang-desktop & co.

#21 2009-01-27 03:33:24

Re: Flawed approach in crunchbang-desktop & co.

#22 2009-01-27 10:23:38

Re: Flawed approach in crunchbang-desktop & co.

Board footer