# 00:07:54 |
codyps |
quits : Ping timeout: 246 seconds |
# 01:13:02 |
bhundven |
quits : Changing host |
# 01:13:02 |
bhundven |
joins #crosstool-ng |
# 01:24:03 |
devcoder |
quits : Quit: devcoder |
# 01:24:03 |
devcoder_ |
is now known as: devcoder |
# 03:49:10 |
bhundven |
is now known as: bhundven|afk |
# 06:08:19 |
alan_o |
quits : Remote host closed the connection |
# 07:53:43 |
bhundven|afk |
quits : Ping timeout: 256 seconds |
# 08:26:37 |
smartin |
joins #crosstool-ng |
# 10:12:32 |
someone1 |
quits : Remote host closed the connection |
# 10:37:32 |
Net147 |
joins #crosstool-ng |
# 12:22:45 |
bhundven|afk |
joins #crosstool-ng |
# 12:23:04 |
bhundven|afk |
quits : Client Quit |
# 12:33:26 |
ccole |
quits : Read error: Connection reset by peer |
# 12:34:13 |
Net147 |
quits : Quit: Want to be different? Try HydraIRC -> http://www.hydrairc.com <- |
# 12:51:16 |
devcoder |
quits : Quit: devcoder |
# 12:51:54 |
devcoder |
joins #crosstool-ng |
# 14:20:36 |
y_morin |
joins #crosstool-ng |
# 15:31:08 |
bhundven |
joins #crosstool-ng |
# 15:54:59 |
bhundven |
ger, I dorked v3 too |
# 15:55:16 |
bhundven |
I'll make sure this one works before I submit the next one :) |
# 16:12:18 |
y_morin |
bhundven: Hehe! Cafeine-deprived, too? ;-) |
# 16:12:25 |
bhundven |
gesh, I guess |
# 16:14:50 |
y_morin |
bhundven: I've pushed my rude wording removal patch. Don't bother rebasing your patchset, I'll do locally. |
# 16:15:10 |
bhundven |
hrm? |
# 16:15:22 |
bhundven |
it's easy to fix |
# 16:15:49 |
y_morin |
bhundven: as you wish ;-) |
# 16:17:09 |
bhundven |
lol |
# 16:22:38 |
bhundven |
everytime I look at gcc.sh I find another mistake I've made |
# 16:23:25 |
bhundven |
I if'd out unziping ecj extraction |
# 16:23:30 |
bhundven |
eh |
# 16:34:37 |
bhundven |
y_morin: question about the CC_VERSION |
# 16:34:52 |
bhundven |
if I use branches/gcc-4_7-branch |
# 16:34:55 |
bhundven |
for instance |
# 16:35:00 |
bhundven |
it becomes: gcc-branches_gcc-4_7-branch |
# 16:35:28 |
y_morin |
So? |
# 16:35:44 |
bhundven |
ok, just check to make sure we both expected that |
# 16:35:51 |
bhundven |
*checking |
# 16:36:31 |
y_morin |
bhundven: I agree it's not pretty, but that's a peculiar case: building with SVN |
# 16:36:38 |
bhundven |
true |
# 16:37:08 |
codyps |
joins #crosstool-ng |
# 16:37:10 |
bhundven |
I'm also playing with a fetch wrapper like I was talking about |
# 16:37:47 |
y_morin |
bhundven: there's another issue with SVN as it is coded today: the version string does not hint at the revision number, so ct-ng will hapilly use a branch tarball if it already exists, and will not get from SVN. |
# 16:37:57 |
bhundven |
eh |
# 16:38:00 |
bhundven |
right |
# 16:38:12 |
y_morin |
(but not special to your code in gcc.sh, there's the same issue in eglibc) |
# 16:38:20 |
bhundven |
right, I was just going to say that |
# 16:38:38 |
bhundven |
ok |
# 16:38:46 |
y_morin |
bhundven: I have no idealy solution, except to append the revision as part of the version string... |
# 16:38:55 |
y_morin |
*ideal |
# 16:38:58 |
bhundven |
right |
# 16:39:13 |
bhundven |
since we don't currently export |
# 16:39:14 |
bhundven |
er |
# 16:39:19 |
bhundven |
sorry, checkout |
# 16:39:32 |
y_morin |
;-) |
# 16:39:49 |
bhundven |
non-ideally, we could delete the tarball if checkout is not selected |
# 16:41:34 |
pradeepto |
hi, the bootstrap script checks for autoconf 2.67 using AC_PREREQ, unfortunately, I have a box that has it installed in some non standard path. A older version is installed in /usr/bin though. Is there a way to make this work? |
# 16:42:00 |
bhundven |
pradeepto: arch linux? |
# 16:42:16 |
pradeepto |
mac os x |
# 16:42:21 |
bhundven |
even funner |
# 16:42:26 |
pradeepto |
tell me about it |
# 16:43:06 |
bhundven |
which version of ct-ng are you using? |
# 16:43:39 |
pradeepto |
brew installs it somewhere randomly, and because I don't have the latest and greatest Xcode, brew refuses to symlink the latest auto* that I installed. |
# 16:43:44 |
bhundven |
hey, y_morin, topic is out of date again :) |
# 16:43:55 |
pradeepto |
bhundven: one from the mercurial repo |
# 16:44:46 |
bhundven |
pradeepto: right, you need to add the brew bin directory to your path before building crosstool-ng |
# 16:45:09 |
bhundven |
my mac is currently running my linux hd, not my mac hd, so I can't remember the path |
# 16:45:17 |
bhundven |
:p |
# 16:45:37 |
pradeepto |
ah, i should have thought of that, somehow i thought it would be more than just the PATH foo. |
# 16:46:25 |
bhundven |
brew docs say: export PATH="/usr/local/bin:$PATH:/usr/local/sbin" |
# 16:46:46 |
pradeepto |
brew docs for? |
# 16:46:51 |
bhundven |
homebrew |
# 16:46:58 |
bhundven |
their documentation |
# 16:47:00 |
bhundven |
... |
# 16:47:00 |
pradeepto |
ah ok, brew docs it self. |
# 16:47:01 |
pradeepto |
ok ok |
# 16:47:03 |
bhundven |
:) |
# 16:47:21 |
bhundven |
it'd probably be easier to add that to your ~/.profile |
# 16:49:57 |
bhundven |
er... ~/.bashrc |
# 16:50:04 |
pradeepto |
right, it unfortunately won't be in /usr/local/bin anyway |
# 16:50:07 |
pradeepto |
that is the problem |
# 16:50:08 |
bhundven |
I don't remember if mac has a ~/.profile |
# 16:50:16 |
pradeepto |
old Xcode blah blah |
# 16:50:26 |
bhundven |
well, the path to your brew/bin |
# 16:50:29 |
pradeepto |
so i am adding the PATH to cellar/foo/bin |
# 16:51:02 |
bhundven |
so, /usr/local/bin doesn't have symlinks into cellar? |
# 16:51:30 |
pradeepto |
hopes, not for automake and autoconf that I installed, rest are there. |
# 16:51:49 |
y_morin |
bhundven: done! ;-) |
# 16:51:50 |
bhundven |
:D |
# 16:51:51 |
pradeepto |
â¦This formula is keg-only, so it was not symlinked into /usr/local. |
# 16:51:51 |
pradeepto |
Xcode (up to and including 4.2) provides (a rather old) Autoconf. <-- I had got that message. |
# 16:51:53 |
pradeepto |
bhundven: ^ |
# 16:52:25 |
bhundven |
ah |
# 16:52:26 |
bhundven |
ok |
# 16:53:11 |
y_morin |
pradeepto: as bhundven said, be sure to set your PATH so that your 'local' autoconf takes precendence over the 'system' one, and you're done. |
# 16:53:32 |
bhundven |
or use the optional autoconf/automake/m4 ct-ng builds |
# 16:53:39 |
pradeepto |
yes, that is what I am doing. |
# 16:54:17 |
y_morin |
bhundven: but that's too late, because ct-ng builds its own to build the components, not to build itself! ;-) |
# 16:54:37 |
bhundven |
oh |
# 16:54:38 |
bhundven |
hehe |
# 16:54:40 |
bhundven |
true |
# 16:54:44 |
pradeepto |
hmmm, what kind of tools does mac os x come with? :O |
# 16:54:47 |
pradeepto |
checking whether sed understands -r -i -e... no |
# 16:54:47 |
pradeepto |
configure: error: |
# 16:54:49 |
y_morin |
bhundven: I understand that pradeepto's issue is with ./bootstrap from ct-ng, as he's using the repository. |
# 16:54:49 |
pradeepto |
bwahhahah! |
# 16:54:54 |
bhundven |
right |
# 16:54:59 |
bhundven |
:$ |
# 16:55:00 |
pradeepto |
this worked so nicely on kubuntu only few hours back |
# 16:55:05 |
bhundven |
runs off to make more coffee |
# 16:55:13 |
pradeepto |
y_morin: yes, not a good idea? |
# 16:55:38 |
y_morin |
pradeepto: you need to tell configure where to find the GNU equivalents: --with-sed=gsed and so on... |
# 16:55:40 |
pradeepto |
btw, thanks for such an awesome tool :) |
# 16:55:48 |
y_morin |
pradeepto: yes, you can use the repository, of course! |
# 16:56:18 |
y_morin |
pradeepto: but if that's your first time, maybe using a release first will cut down some issues (maybe!). |
# 16:56:24 |
y_morin |
pradeepto: Cheers! |
# 16:56:56 |
pradeepto |
y_morin: indeed it is, it is second time, first time went well on a linux box. I thought I will try it on os x |
# 16:57:48 |
pradeepto |
y_morin: ok, i will use a release version for now, at least on mac |
# 16:57:49 |
y_morin |
pradeepto: run ./configure --help : it wil tell you the options to pass to specify the path to the required tools, if they can not be found |
# 16:58:15 |
y_morin |
pradeepto: even with a release, you'll have the same issues wrt to the GNYU tools being required |
# 16:58:19 |
y_morin |
*GNU |
# 16:58:35 |
pradeepto |
right, i didn't know sed in os x wasn't the GNU one |
# 16:58:54 |
y_morin |
pradeepto: in MacOS-X |
# 16:59:00 |
pradeepto |
wait, of course ,it is the freebsd on, is it? |
# 16:59:12 |
y_morin |
these tools are most probably coming from the BSD land. |
# 16:59:17 |
y_morin |
pradeepto: yep! |
# 16:59:51 |
y_morin |
pradeepto: http://code.bulix.org/dup47k-82011 <-- the list of required GNU tools and the options to specify them. |
# 16:59:59 |
pradeepto |
checking |
# 17:00:26 |
y_morin |
pradeepto: there's a doc (in docs/) on how to use ct-ng on MacOS-X. |
# 17:01:17 |
y_morin |
pradeepto: http://crosstool-ng.org/hg/crosstool-ng/file/c87cc34319ab/docs/C%20-%20Misc.%20tutorials.txt |
# 17:01:49 |
pradeepto |
damn, my bad, I did read the installation.txt but not the tutorials one |
# 17:01:51 |
pradeepto |
will check |
# 17:05:02 |
pradeepto |
y_morin: how can I send you patches? |
# 17:05:23 |
y_morin |
pradeepto: there's a doc (in docs) about how to contribute! ;-) |
# 17:05:26 |
y_morin |
Hehe! |
# 17:05:35 |
bhundven |
hugs the documentation |
# 17:05:44 |
pradeepto |
pfft! you have though about it all, haven't you! ;) |
# 17:06:11 |
y_morin |
pradeepto: if I had thought oabout it all, there would be no need to send patches! ;-) |
# 17:06:46 |
pradeepto |
greped for "contribution" on the website and found nothing |
# 17:07:06 |
pradeepto |
had a good laugh though, thanks to you Contacts table's last row |
# 17:07:08 |
pradeepto |
:P |
# 17:07:27 |
y_morin |
pradeepto: look at file '7 - Contributing to crosstool-NG.txt' for a quicky, and 'C - Misc. tutorials.txt' for an in-depth tuto. |
# 17:07:37 |
pradeepto |
yes, i am on it |
# 17:08:14 |
y_morin |
pradeepto: well, I found that womewhere (can't remember where), and thought it be nice to use! ;-) |
# 17:08:21 |
y_morin |
*somewhere |
# 17:08:30 |
bhundven |
http://crosstool-ng.org/hg/crosstool-ng/file/987ff9768880/docs/C%20-%20Misc.%20tutorials.txt#l112 |
# 17:08:45 |
bhundven |
is where I'd suggest starting |
# 17:09:02 |
bhundven |
because mq is so friggin handy |
# 17:15:40 |
y_morin |
goes off for a well-earned beer at the end of the day! |
# 17:15:58 |
pradeepto |
cheers! |
# 17:15:59 |
y_morin |
See ya, aligator! |
# 17:20:12 |
bhundven |
enjoy |
# 17:27:28 |
bhundven |
alright, this revision is going much better |
# 18:41:53 |
ctngbot |
joins #crosstool-ng |
# 18:48:04 |
bhundven |
pradeepto: yeay |
# 18:50:31 |
bhundven |
gets his ppc setup back up and running |
# 18:57:47 |
ribamar |
joins #crosstool-ng |
# 19:00:22 |
ribamar |
hi there..... has anyone built a toolchain for raspberry pi? |
# 19:33:14 |
bhundven |
ribamar: I'm not sure. I would suggest following the raspberry pi documentation. I plan on getting on of these soon, and hopefully add an official defconfig for ct-ng. |
# 19:34:02 |
ribamar |
bhundven, i followed http://www.bootc.net/archives/2012/05/26/how-to-build-a-cross-compiler-for-your-raspberry-pi/ and it worked for me |
# 19:34:55 |
bhundven |
ok |
# 19:35:17 |
bhundven |
so you answered your own question |
# 19:35:21 |
bhundven |
You have ;) |
# 19:35:44 |
ribamar |
but it's weird that the generated gcc dumps: arm-unknown-linux-gnueabi-gcc -dumpmachine |
# 19:35:45 |
ribamar |
arm-unknown-linux-gnueabi |
# 19:36:04 |
bhundven |
that is right |
# 19:36:17 |
ribamar |
whilst the raspbian gcc dumps: arm-linux-gnueabihf |
# 19:36:57 |
ribamar |
and so, i'm getting trouble compiling bigger projects (qt, in my case) |
# 19:37:29 |
bhundven |
what fails in qt? |
# 19:37:50 |
ribamar |
I wandering if there is any way to change the value CT_TARGET from arm-unknown-linux-gnueabi to arm-linux-gnueabihf |
# 19:38:41 |
bhundven |
you could try, but ymmv. they probably have some custom patches for that target. i've never heard of gnueabihf |
# 19:39:22 |
ribamar |
they = who ? raspiberry foundation? |
# 19:39:46 |
ccole |
joins #crosstool-ng |
# 19:41:09 |
bhundven |
I don't know. I can't speak for them. |
# 19:41:11 |
ribamar |
bhundven, but do you know how/where to change CT_TARGET ? do you believe that there are chances making that chage to work ? |
# 19:41:31 |
bhundven |
... |
# 19:41:33 |
bhundven |
I don't know. I can't speak for them. |
# 19:41:53 |
bhundven |
maybe someone else here has experience with that? If so, they are better to help you then me. |
# 19:42:16 |
bhundven |
or try asking on their forum |
# 19:42:23 |
ribamar |
ok, thanks anyway |
# 19:44:01 |
bhundven |
oh, hf, hardfloat |
# 19:44:15 |
bhundven |
-march=armv6 -mfpu=vfp -mfloat-abi=hard |
# 19:45:14 |
bhundven |
http://www.raspberrypi.org/phpBB3/viewtopic.php?f=66&t=7413 |
# 19:45:56 |
bhundven |
try also searching google(as I did) with: site:www.raspberrypi.org toolchain crosstool |
# 19:47:27 |
ribamar |
bhundven, to reach the point I already have done a deep search for it.... |
# 19:47:48 |
bhundven |
then you know more then I do. |
# 19:48:08 |
ribamar |
thanks anyway for trying to help... |
# 19:49:00 |
bhundven |
you never did say what failed in qt |
# 19:50:14 |
bhundven |
also, not everyone in this irc chatroom is awake |
# 19:50:25 |
bhundven |
try sending the same infromation we've talked about to the mailing list |
# 19:51:08 |
ribamar |
https://gitorious.org/cross-compile-tools/cross-compile-tools won't create correct links if gcc -dumpmachine dumps the wrong arch. |
# 19:53:43 |
bhundven |
so set the CT_TARGET to have the right machine string |
# 20:05:26 |
devcoder |
joins #crosstool-ng |
# 20:09:02 |
ribamar |
bhundven, yeah, i'm looking for someone who knows where to change the target name (CT_TARGET). i couldn't find it in ct-ng menuconfig (wondering if it is possible) |
# 20:09:21 |
ribamar |
bhundven, but don't care, i'll ask here another time.... |
# 20:28:18 |
y_morin |
ribamar: still here? |
# 20:30:20 |
y_morin |
ribamar: basically, -gnueabihf and -gnueabi are exactly the same: |
# 20:30:54 |
y_morin |
ribamar: - both use the EABI (eg. calling convention, structure layout...) |
# 20:31:25 |
y_morin |
ribamar: - and both use the same floating point ABI, that is both use software FP 'layout' |
# 20:32:02 |
y_morin |
ribamar: it's just that -gnueabohf uses hardware floating point instructions, while -gnueabi uses only software emulation. |
# 20:32:25 |
y_morin |
--> -gnueabihf (not gnueabohf!) |
# 20:32:51 |
Juv1228 |
quits : Remote host closed the connection |
# 20:32:56 |
y_morin |
ribamar: so they are both interoperable (ie. -gnueabihf code can call -gnueabi, and vice-versa) |
# 20:33:43 |
y_morin |
ribamar: crosstool-NG does *not* distinguish both cases, so hard FP and software FP both end up with -gnueabi |
# 20:34:01 |
y_morin |
ribamar: Note that for ARM, there are three kinds of floating point ABI; |
# 20:34:26 |
y_morin |
ribamar: - software ABI with software emulation |
# 20:34:38 |
y_morin |
ribamar: - software ABI with hardware instructions |
# 20:34:49 |
y_morin |
ribamar: - hardware ABI with hardware instructions |
# 20:35:14 |
y_morin |
ribamar: of course, -gnueabi is the first, while -gnueabihf is the second, they are interoperable. |
# 20:35:33 |
y_morin |
ribamar: the third, as it uses a different ABI, can not interoperate with the two others. |
# 20:35:43 |
y_morin |
And I am done for now, so you can speak! ;-) |
# 20:36:06 |
y_morin |
bhundven: Seen the explanations above about -gnueabi vs. -gnueabihf |
# 20:41:05 |
ChanServ |
joins #crosstool-ng |
# 20:52:47 |
ribamar |
y_morin, i just came back (still reading) |
# 21:01:47 |
y_morin |
ribamar: was that informative? |
# 21:01:52 |
ribamar |
y_morin, thanks for the explanation. my point is: even if they are interoperable, my problem is not exactly with interoperability (I do succeed in cross-compiling). however, ct-NG doesn't distinguish -gnueabi and -gnueabihf, some build systems |
# 21:01:57 |
ribamar |
will be mislead |
# 21:02:16 |
ribamar |
by the arm-unknown-gcc -dumpmachine |
# 21:02:36 |
y_morin |
ribamar: that should not be a problem. Just tell those build systems (which ones?) that the tuple is arm-unknown-linux-gnueabi |
# 21:03:26 |
y_morin |
ribamar: I'll look to see if it is easy to add hf to the tuple in a little while (other duties right now). |
# 21:04:24 |
y_morin |
ribamar: the problem with -gnueabihf is that it is quite new (less than a year iirc), so some packages may refuse to build because they bundle old config.{syb,guess} which do not know about -gnueabihf. |
# 21:05:05 |
y_morin |
bhundven: I saw your gcc svn patch. Will look in a little while (banging on some other stuff atn). |
# 21:05:08 |
y_morin |
*atm |
# 21:08:15 |
ribamar |
quits : Ping timeout: 272 seconds |
# 21:27:12 |
devcoder |
quits : Quit: devcoder |
# 21:48:42 |
bhundven |
np |
# 21:52:00 |
bhundven |
I'm sure it's getting late for you anyways. |
# 21:53:42 |
smartin |
quits : Quit: leaving |
# 21:54:17 |
y_morin |
bhundven: almost midnight, that is, about 1-2 hours left! :-) |
# 22:02:38 |
bhundven |
got my freescale ppc box plugged in again |
# 22:03:02 |
bhundven |
its been a while... |
# 22:18:15 |
ribamar |
joins #crosstool-ng |
# 22:18:34 |
devcoder |
joins #crosstool-ng |
# 22:19:18 |
mnt_real |
quits : Ping timeout: 276 seconds |
# 22:21:44 |
mnt_real |
joins #crosstool-ng |
# 22:31:45 |
bhundven |
y_morin: I was looking at the CT_GetSource wrapper I was talking about |
# 22:32:41 |
y_morin |
bhundven: yep? |
# 22:32:44 |
bhundven |
and it looks like we need to normalize the way GetFile, GetCVS, GetSVN, and GetGIT work. for instance, they don't all return the same way |
# 22:33:21 |
bhundven |
maybe make their arguments named arguments like the build wrapper in gcc.sh |
# 22:34:01 |
y_morin |
bhundven: GetFile was supposed to be the real frontend, and the SVN/CVS/GIT only backends that should not be called directly. |
# 22:34:08 |
bhundven |
hmm |
# 22:34:14 |
y_morin |
Turned out it's not how it turned in the end. |
# 22:34:29 |
bhundven |
yea |
# 22:34:36 |
y_morin |
They all ended up being frontends. |
# 22:34:41 |
bhundven |
I was thinking of something like: |
# 22:34:51 |
bhundven |
declare -A url |
# 22:35:16 |
bhundven |
url[git]=https://url1... |
# 22:35:26 |
bhundven |
url[git]+=https://url2... |
# 22:35:36 |
bhundven |
url[svn]+=https://svn1... |
# 22:35:45 |
bhundven |
url[file]+=https://www... |
# 22:35:49 |
y_morin |
bhundven: yep, that's what I was thinking too. |
# 22:35:58 |
bhundven |
then have the wrapper pass to the right getter |
# 22:36:10 |
bhundven |
depending on the method |
# 22:36:12 |
y_morin |
bhundven: but bash only handles one-dimensional arrays |
# 22:36:19 |
bhundven |
in 4.0 |
# 22:36:26 |
bhundven |
-A is associative arrays |
# 22:36:32 |
bhundven |
opposed to -a |
# 22:36:51 |
y_morin |
bhundven: I know, but still, associative in one dimension. |
# 22:37:17 |
bhundven |
fine, for the url and method, only need the key=method, value=url. we can pass the rest as named arguments |
# 22:37:25 |
y_morin |
bhundven: foo['a']=123; foo['a']+=456; echo ${foo['a']}" --> '123456' |
# 22:37:42 |
bhundven |
oh, I meant |
# 22:38:13 |
y_morin |
bhundven: so we can have only one url of any 'kind' |
# 22:38:22 |
bhundven |
url[method]=( "oneurl" ), url[method2]=( "anotherurl" ) |
# 22:38:37 |
bhundven |
hmm |
# 22:38:50 |
y_morin |
bhundven: for some components, it's not possible, as we need to check many URLs to find the sources: |
# 22:39:15 |
y_morin |
for example, binutils snapshots are on kernel.org, while releases are on gnu.org |
# 22:39:17 |
alan_o |
joins #crosstool-ng |
# 22:39:22 |
bhundven |
ah |
# 22:39:23 |
bhundven |
ok |
# 22:39:37 |
bhundven |
hehe |
# 22:39:56 |
bhundven |
just use variable names like: release[method] or snapshot[method] |
# 22:39:57 |
y_morin |
So, the caller (eg. binutils' do_get() ) must be able to pass many URLs. |
# 22:39:58 |
bhundven |
;) |
# 22:40:10 |
bhundven |
depending on if you are getting a snapshot or release |
# 22:40:32 |
y_morin |
bhundven: BUT! If we do something like (going to some pastebin...) |
# 22:40:33 |
bhundven |
the wrapper shouldn't care about the name of the array |
# 22:40:38 |
bhundven |
k |
# 22:42:45 |
bhundven |
oh |
# 22:42:47 |
bhundven |
hmm |
# 22:42:48 |
bhundven |
bryan@umactu:~$ foo["git"]+=( "hello1" ) |
# 22:42:48 |
bhundven |
-bash: foo["git"]: cannot assign list to array member |
# 22:42:50 |
bhundven |
ok |
# 22:43:04 |
y_morin |
bhundven: http://code.bulix.org/wbsba7-82014 |
# 22:43:37 |
bhundven |
oh |
# 22:43:57 |
bhundven |
and we strip the method from the beginning of the url... if it is there, otherwise expect it to be a file |
# 22:44:00 |
bhundven |
? |
# 22:44:06 |
y_morin |
We'd still need to pass some extra info (eg. svn export/checkout and revision, git ref...) |
# 22:44:11 |
bhundven |
sure |
# 22:44:33 |
bhundven |
I knew there would be some dumb limitation to bash associative arrays |
# 22:44:39 |
bhundven |
eh |
# 22:44:50 |
y_morin |
bhundven: yep, the caller just says what he wants, then CT_GetFile parses the URL (which is now a URI) scheme and acts appropriately. |
# 22:44:59 |
bhundven |
right |
# 22:45:02 |
bhundven |
that makes sense |
# 22:45:35 |
bhundven |
just for technicallity reasons: file://foo/bar/buz |
# 22:45:41 |
bhundven |
should be: file:///foo/bar/buz |
# 22:45:43 |
bhundven |
;) |
# 22:45:55 |
bhundven |
I know it was an example |
# 22:45:56 |
y_morin |
Something like: scheme://path/to/rssource[@reference] |
# 22:46:09 |
y_morin |
bhundven: yep! |
# 22:46:09 |
bhundven |
well |
# 22:46:22 |
bhundven |
but you have to be careful what you use as the demark for reference |
# 22:46:30 |
bhundven |
http://user:pass@url... |
# 22:46:32 |
y_morin |
Which would give for git: git://server/repo@cset |
# 22:47:00 |
y_morin |
bhundven: no need to support that! It will not be user-defined URIs! |
# 22:47:23 |
y_morin |
We're not speaking about letting ythe user point to his/her own repositories, are we? |
# 22:47:27 |
bhundven |
oh, thats where I was going after the wrapper, was allowing the user to specify maybe their own git repository |
# 22:47:33 |
y_morin |
Ah... |
# 22:47:33 |
bhundven |
or svn |
# 22:47:54 |
y_morin |
Not too sure we should go that route... |
# 22:48:04 |
bhundven |
I have my own repos of stuff and would like to build stock crosstool-ng with my custom hacked gcc or glibc without having to muck with patches |
# 22:48:37 |
y_morin |
bhundven: I understand. That's why I originally added the "custom patchset" option, so one could use local changes |
# 22:48:57 |
bhundven |
yea, but if I wanted to use my own patches, I'd start using darcs |
# 22:48:59 |
y_morin |
I never really thought about using a local repository. |
# 22:49:07 |
bhundven |
yikes |
# 22:49:17 |
bhundven |
I'd rather use mercurial or git |
# 22:49:49 |
y_morin |
bhundven: I see your point. |
# 22:50:18 |
y_morin |
But, crosstool-Ng was not designed with the 'core component devlopper' as a target for ct-ng. |
# 22:51:11 |
y_morin |
Initially, the target audience was people that wanted to build toolchain on a reproducible way, without having to care about all the nitty-gritty details |
# 22:51:40 |
y_morin |
bhundven: Not to say that I am against integrating such a change! ;-) |
# 22:52:05 |
bhundven |
sorry, I had to go wrangle my daughter |
# 22:52:11 |
y_morin |
bhundven: but if we go that route, better be sure we iron the issues first! |
# 22:52:16 |
bhundven |
who had cake all over her |
# 22:52:22 |
bhundven |
yes |
# 22:52:27 |
y_morin |
bhundven: of course real-life takes precedence! Familly even more so! |
# 22:52:31 |
bhundven |
so lets forget about that idea for now |
# 22:52:52 |
y_morin |
bhundven: no, but lets think about it straight. |
# 22:52:57 |
bhundven |
sure |
# 22:53:07 |
bhundven |
I was thinkging lets just work on the wrapper first |
# 22:53:13 |
bhundven |
and everything it depends on |
# 22:53:22 |
y_morin |
bhundven: right. |
# 22:53:37 |
bhundven |
because once that is straight, everything else will fall right into place |
# 22:53:54 |
y_morin |
Then it'll be realtively easy to just add an option in the menuconfig to offer some kind of "local repository" option |
# 22:54:01 |
bhundven |
yup |
# 22:54:17 |
y_morin |
Good. We're back on track! ;-) |
# 22:54:21 |
bhundven |
:D |
# 22:54:35 |
y_morin |
Damn nicotine-addiction... BRB... |
# 22:54:50 |
bhundven |
srsly. I have to wait for mom before I can |
# 22:55:35 |
bhundven |
but, its better that way. down to 2 a day |
# 22:59:02 |
y_morin |
:-) |
# 23:00:47 |
y_morin |
wishes he'd be able to switch contest as easily as a CPU does, to be able to multi-task maore easily... :-/ |
# 23:00:53 |
y_morin |
*context |
# 23:01:03 |
y_morin |
*more |
# 23:01:14 |
y_morin |
is need of some sleep, too, it seems! :-) |
# 23:01:33 |
y_morin |
Gah.. OK, leyt's call it a day |
# 23:01:42 |
bhundven |
k, g'night |
# 23:01:52 |
y_morin |
Bye! |
# 23:01:59 |
y_morin |
quits : Quit: Nighty Night! |