<rafa>
wpwrak: dd.. no idea.. if mkswap does that then it is something new for me
<rafa>
:)
<rafa>
(and i am an old admin :D)
<wpwrak>
rafa: (advertizing) setting the prompt to "Jlime". good product placement :)
<wolfspraul>
xiangfu: maybe you have seen the new config.miminal and config.xbboot style?
<wolfspraul>
I have experimented a bit with make oldconfig in OpenWrt, and so far it works quite well
<wolfspraul>
OpenWrt has no silentoldconfig so the way to get the config files picked up is yes ""|make oldconfig>/dev/null
<wolfspraul>
maybe they can introduce a silentoldconfig later if they want
<wolfspraul>
I will now commit the new, much shorter, config.full_system
<wolfspraul>
only 8 KB instead of 110 KB
<wolfspraul>
what we still need to do over time is to cleanup config.full_system inside - order apps and libraries in meaningful groups, for some document what they are, etc.
<wolfspraul>
I am currently running a full build with the new, shorter config.full_system, but will commit it already
<wolfspraul>
check it out and let me know what you think...
<xiangfu>
wolfspraul: I also try 'make oldconfig' today.
<xiangfu>
I am test the your commit now. :)
<wolfspraul>
it works quite well I think
<wolfspraul>
it seems to ignore comment lines (though gives warnings on empty lines :-))
<wolfspraul>
no silentoldconfig unfortunately, but once you know the yes "" thing it works quite well
<wolfspraul>
I am currently running a test build with the config.full_system I just committed, should finish in a few hours.
<wolfspraul>
if it produces a bootable image, I will then spend a few hours to reorder/group the config file in meaningful groups
<wolfspraul>
at that point we hopefully have a basis that people can easily and quickly add meaningful apps and libraries too
<wolfspraul>
that's my plan roughly...
<wolfspraul>
then I still need an image for my reflashing job which starts tomorrow. Any progress on the sdl/qt bugs?
<xiangfu>
no.
<xiangfu>
hard to find.  I search a lot. but I can  not find a answer.
<wolfspraul>
well I think we just need classical debugging tools. debug symbols, gdb/gdbserver, etc.
<wolfspraul>
I created a config.minimal, did you see that?
<xiangfu>
wolfspraul: I will remove the 'ubi-utils' in openwrt-package.git.  which is the same 'mtd-utils' in openwrt.
<xiangfu>
yes. I saw the config.minimal
<wolfspraul>
I'm not sure whether it's the same. is it really?
<wolfspraul>
I thought there are different tools in ubi-utils that are not in mtd-utils. Do they point to the same upstream sources?
<wolfspraul>
for config.minimal, my plan is to only add one app, say NanoMap, or imgv, but turn all debug symbols on. the image is then still relatively quick to build.
<wolfspraul>
that way hopefully it's possible to zoom in on bugs quickly
<xiangfu>
mtd and ubi point to the same upstream sources.
<wolfspraul>
oh
<wolfspraul>
hmm
<wolfspraul>
is ubiformat in mtd-utils?
<xiangfu>
yes
<wolfspraul>
well then, I think when we created ubi-utils there was some problem with mtd-utils
<wolfspraul>
maybe that has been resolved since - so yeah, let's delete ubi-utils
<wolfspraul>
do you know how we can display the diff between tracking_backfire and openwrt backfire, and between tracking_trunk and openwrt trunk?
<xiangfu>
hmm.. we need add two remote git url for diff. for example tracking_trunk...openwrt_trunk
<xiangfu>
4. git diff  openwrt_backfire...tracking_backfire
<xiangfu>
since we always use 'git merge' not 'git rebase'Â Â it's very hard to see what we have changed.
<xiangfu>
wolfspraul: "this config file does not work on tracking_trunk branch - why?" there are some commit for zImage and xbboot only in tracking_backfire branch.
<xiangfu>
not into upstream.
<wolfspraul>
xiangfu: why are we not rebasing the tracking_ branches?
<wolfspraul>
if you say 'very hard to see' then I just interpret that as 'impossible' :-)
<wolfspraul>
and that means that we need to run a diff -ru from outside on the entire tree (that always works), and/or use rebase instead of merge
<wolfspraul>
the goal of the tracking_ branches in the openwrt-xburst project is to be as small as possible, so if the diff is hard to see we will not reach that goal, for sure
<wolfspraul>
over time more and more crap will accumulate there
<wolfspraul>
what do you think we should do?
<xiangfu>
how about we only change the those folder:
<xiangfu>
target/linux/xburst/
<xiangfu>
package/uboot-xburst/
<xiangfu>
data/
<xiangfu>
scripts/build
<xiangfu>
scripts/reflash_ben.sh
<xiangfu>
if someone think he/she want change the openwrt system files.  he/she should change it in his personal branch. or send it to upstream.
<xiangfu>
if someone want change
<xiangfu>
wolfspraul: we need create a new branch for this. it's impossible to rebase the tracking_backfire branch.
<wolfspraul>
we should definitely wait until October when Mirko is back to get his advice. in the meantime I just though I look at the diff a little to understand what we are talking about.
<wolfspraul>
do you think we only made changes in those directories you mentioned? plus the build and reflash_ben.sh scripts?
<xiangfu>
wolfspraul: yes.
<wolfspraul>
can we move the uboot-xburst package into openwrt-packages?
<xiangfu>
yes. we can but it's not good. if we move it to openwrt-package. then upstream must add our openwrt-packages to feeds.conf
<kristianpaul>
Good Morning :)
<xiangfu>
since the u-boot is core part of make ben image. it's better to add it in package/ folder.
<kyak>
xiangfu: do you know how to create fonts (pnm files) for setfont2? i tryed playing with gimp and netpbm, but can't produce a font -\
<wolfspraul>
xiangfu: oh so the package/uboot-xburst folder is also in the upstream repository?
<wolfspraul>
and target/linux/xburst as well I guess
<xiangfu>
wolfspraul: yes. both are in the upstream repository
<kyak>
wolfspraul: looking at config.full_system: do we need CONFIG_BUSYBOX_CONFIG_HTTPD=y? also we might want to disable CONFIG_BUSYBOX_CONFIG_SETFONT, because setfont is provided by kbd, and busybox uses some other file format...
<wolfspraul>
you get that 'hung up' message when you try to push into the read-only url git://projects.qi-hardware.com/openwrt-xburst.git
<kristianpaul>
hmm
<xiangfu>
kyak: sorry. I don't know. we need ask "Unclouded"
<kristianpaul>
i just was following instructions to get the update about branch renaming
<wolfspraul>
kyak: I like the idea of having a http server because you can connect from your notebook and have a 'big screen', for example as a development environment
<kyak>
xiangfu: ok, no problem :)
<wolfspraul>
for the busybox setfont - sure if you think it's redundant, then out with it!
<kyak>
wolfspraul: hm, i thought it's SSH server?
<wolfspraul>
that's why I started working towards a cleaner config, so we can finally zoom in on a collection of valuable stuff
<wolfspraul>
httpd is a ssh server? well, don't know. dropbear is the ssh server.
<kyak>
no-no.. "big screen" as a development environment is an SSH server :0
<wolfspraul>
kyak: if you think the busybox_setfont is redundant and not needed - don't hesitate and just remove it.
<kyak>
what do you expect to do with httpd ?
<kristianpaul>
UI?
<kristianpaul>
upload music?
<xiangfu>
kristianpaul: whey you want "push origin master:tracking_trunk" ??
<wolfspraul>
kyak: ah sure, yes. but imagine you have some php or python or whatever scripts installed that let you access the nanonote from another computer. nice applications are imaginable.
<kristianpaul>
and other stuff may be?
<wolfspraul>
the browser is just a very ubiquitous app...
<kristianpaul>
xiangfu: no, i'm reading now is no cense
<kyak>
hm, ok.. but i can't imagine any use of httpd, when you have sshd
<kyak>
kristianpaul: what do you mean "UI"?
<wolfspraul>
people may only have, or be more familiar, with a browser on the other side
<wolfspraul>
anyway I vote to leave the httpd in
<rafa>
wpwrak: that is the prompt of the shell ;).. a lot nicer than root@alongwordwhenyouhavejust320x240screen ;-))
<kristianpaul>
xiangfu: ahh sorry i was reading wrong part  of your mail :p
<kristianpaul>
kyak: User Interface
<xiangfu>
I was usually think put the nanonote manual  web page to nanonote . then we can http://192.168.254.101 for read the manual . that is an old idea.
<kyak>
wolfspraul: the make oldconfig, will it use default config but take values from .config and override the default value with them?
<kristianpaul>
kyak: i agreee with wolfspraul broweser is friendly and people righ now are capable of attching mails so... :) will be the same with the Ben
<wolfspraul>
yes I think that's how it works
<kristianpaul>
broweser/web browser
<wolfspraul>
the idea with the config files in data/qi_lb60/conf is that you copy them to .config, then you run oldconfig over it
<kyak>
lets say it this way: make oldconfig is using value you provide in .config, and everything else is default?
<kyak>
will it also track dependencies?
<wolfspraul>
I'm actually not sure whether the busybox httpd is good enough, maybe lighttpd is better? but anyway nobody uses it right now so we can start with the lightest approach, which I'm sure is the busybox builtin one...
<kyak>
and why it is named so strange? :)
<wolfspraul>
kyak: what is named strange?
<kyak>
"oldconfig"
<wolfspraul>
yes, I'm pretty sure dependencies work as normal
<wolfspraul>
it's coming from the Linux kernel
<wolfspraul>
I think in general the OpenWrt build system borrows (reuses) a lot of concepts and ideas from the original Linux kernel config system
<wolfspraul>
well, someone more knowledgeable about OpenWrt correct me if that impression is wrong...
<kyak>
yes ,this is so...
<wolfspraul>
kyak: check out config.minimal
<kyak>
hm ok, then it's only a matter of edditing text file now to remove or add some option
<wolfspraul>
if all works well in OpenWrt, it's as if you turn these things on in make menuconfig, and then OpenWrt adds the necessary dependencies
<kyak>
wolfspraul: already did, very cool
<wolfspraul>
yes
<wolfspraul>
the config.xbboot one is quite unstable, I'd say stay away from that unless you know what you are doing. when I remove or add anything there, the resulting initramfs will not boot anymore
<wolfspraul>
but config.minimal and config.full_system should be more robust, we can spend some time to improve config.full_system now
<wolfspraul>
OpenWrt does not seem to have a 'silentoldconfig' option
<kyak>
does Linux kernel have?
<wolfspraul>
so if you just run 'make oldconfig', it will ask you a one-line question for every new package or package/setting that was not in your .config
<wolfspraul>
I think it does, yes.
<kyak>
but your workaround seems ok
<kyak>
echo "yes"
<wolfspraul>
but we can simulate silent with yes ""|make oldconfig > /dev/null
<wolfspraul>
it's not echoing yes, it's echoing 'enter'
<wolfspraul>
so we take the default value OpenWrt suggests
<kyak>
oh right
<wolfspraul>
just run yes ""
<wolfspraul>
:-)
<kristianpaul>
xiangfu: i noticed new stuff when run the reflash script, so i guess the sources for this are in the OpenWrt Release (Backfire)Â Â isnt?
<kyak>
(also now to figure out how to undo the last commit)
<wolfspraul>
don't know. maybe xiangfu can delete the test commit? xiangfu : do you know how to setup git push so that it automatically finds the right remote repository?
<wolfspraul>
kyak: if we are lucky xiangfu is still around :-) otherwise we have to turn on our brains ourselves, and find the answer...
<wolfspraul>
kyak: congratulations for your first openwrt-xburst commit!
<kyak>
when it comes to git, my brain hides somewhere
<kyak>
wolfspraul: thanks!
<kyak>
my usage of git is limited to three commands
<wpwrak>
kyak: git log, git show, and git rebase -i are also worth knowing :)
<kyak>
wpwrak: git rebase is new to me
<wpwrak>
ah, and git reset. undo what git add did. useful if you have two sets of changes, want to put them in separate commits, and made a mistake.
<kyak>
wpwrak: better explain to me what git rebase is :)
<wpwrak>
git rebase -i lets you merge commits (and do a few other things to them). the man page has a good example. only limitation is that you shouldn't have pushed them yet.
<kyak>
man page would be useful, if i knew some basics.. but "Forward-port local commits to the updated upstream head" are just a set of words for me
<kyak>
ok, i see.. it's when i want to merge remote changes into my local branch, but i want my local changes to stay?
<wpwrak>
just ignore the weirdness they talk about :) the interesting bit is in the section INTERACTIVE MODE
<kyak>
is it smart diff? :)
<wpwrak>
it can do that too, yes. but rebase -i is about managing your own commits.
<kyak>
i should check that -i
<wpwrak>
you basically get another layer of "committing", where commits aren't final.
<larsc>
you can rearange, remove, edit and squash commits into each other
<kyak>
this sounds like cool.. one only needs to hope that his project will work continue working after such intensive things :)
<wpwrak>
kyak: if you break something, you can always git rebase -i again ;-))
<larsc>
or reset
<kyak>
ok, but i already pushed
<kyak>
how to revert in this case?
<larsc>
you add another commit reverting your commit
<kristianpaul>
well the one that the update script fetches
<kristianpaul>
i jsut updated my ben today and seemed like a downgrade :(
<xiangfu>
kristianpaul: ok. we make the last image point to 16-Jun-2010 image. since the 2010-08-26 have uClibc 0.9.32 problem.
<kristianpaul>
ahh i see
<xiangfu>
kristianpaul: yes. we downgrade it.
<kristianpaul>
cause i was hoping to see the ogg stuff and nanomap in gmenu too
<kristianpaul>
okay is clear for me now
<xiangfu>
kristianpaul: I found the 'tangogps' is ported to openwrt. we have selected it in config.full_system. just FYI.
<kristianpaul>
xiangfu: yes is i usde in the past
<kristianpaul>
xiangfu: really i dont follow tangogps on openwrt is *really* slow
<kristianpaul>
nanomap save the day in openwrt
<kristianpaul>
for sure :)
<kristianpaul>
i will take care of (remebers the bug in openwrt) about cgps and gpx logger
<kristianpaul>
i said cause i requested it from openwrt
<kristianpaul>
it really will help not just only to Ben
<xiangfu>
sure.
<kristianpaul>
xiangfu: i'm curios as there is a full system config, there is no a minal config, or i may be wrong?
<xiangfu>
kristianpaul: wolfgang have created a config.minimal already.
<kristianpaul>
oh let me push
<kristianpaul>
pull
<kristianpaul>
sorry
<kristianpaul>
hehe great i was thinking that i was alrady made hours ago :)
<kristianpaul>
btw are you using an script in fidelio for the builds?
<kristianpaul>
i it somwhere?
<xiangfu>
kristianpaul: yes.
<kristianpaul>
and is reausble for other users?
<kristianpaul>
good :)
<kristianpaul>
i just want prepare a full and minimal image :)
<xiangfu>
kristianpaul: there is one in git: "tracking_backfire/scripts/build'
<kristianpaul>
okie dokie :)
<xiangfu>
kristianpaul: you can take a look that file you will understand. for use this scripts file. we need copy the config.minimal to .config manually
<kristianpaul>
sure
<xiangfu>
kristianpaul: do you think we should use [yes "" | make oldconfig] instead of [make oldconfig] ??
<xiangfu>
in this 'scripts/build' script file.
<kristianpaul>
hmm is a bit confusing
<kristianpaul>
[ make oldconfig | yes]
<kristianpaul>
echo "Are you brave, dude? [NO/yes]"
<kristianpaul>
lol :)
<xiangfu>
kristianpaul:[make oldconfig | yes] will make the script will never stop. :)
<kristianpaul>
ahh
<kristianpaul>
hehe
<kristianpaul>
i tought was a question not a command
<kristianpaul>
yes is a command i see now
<kristianpaul>
[yes "" | make oldconfig] <<<-- yes i agree
<xiangfu>
yes. I think [yes "" | make oldconfig] is much better.