# 00:50:57 |
djerome |
joins #crosstool-ng |
# 02:15:35 |
djerome |
quits : Remote host closed the connection |
# 03:58:04 |
alan_o |
quits : Remote host closed the connection |
# 04:45:36 |
bhundven |
https://github.com/bhundven/crosstool-ng-build-test |
# 05:12:37 |
bhundven |
wip |
# 07:21:30 |
aalv |
joins #crosstool-ng |
# 13:33:58 |
alan_o |
joins #crosstool-ng |
# 15:47:13 |
y_morin |
joins #crosstool-ng |
# 16:09:05 |
aalv |
quits : Quit: Leaving. |
# 16:36:43 |
doc2 |
joins #crosstool-ng |
# 17:28:40 |
Martell |
joins #crosstool-ng |
# 17:29:57 |
Martell |
hey y_morin :) |
# 17:30:20 |
Martell |
I've seen that cygwin has been changed to supported |
# 17:30:21 |
y_morin |
Martell: Hello! |
# 17:30:27 |
Martell |
on hg |
# 17:30:42 |
Martell |
what about msys? |
# 17:31:17 |
y_morin |
Martell: Well, "cygwin is supported" means "ct-ng can run on cygwin". |
# 17:31:25 |
y_morin |
Martell: Is that what you exoected? |
# 17:31:32 |
y_morin |
*expected |
# 17:31:46 |
Martell |
well ct-ng runs on msys2 |
# 17:32:01 |
Martell |
but I'm having error building with msys as the host |
# 17:32:05 |
y_morin |
Martell: However, ct-ng does not know how to generate toolchains that _target_ cygwin. |
# 17:32:20 |
y_morin |
Martell: Ah, no I have no clue about msys. |
# 17:32:22 |
Martell |
mingw works as a host |
# 17:32:37 |
Martell |
ahh |
# 17:32:49 |
Martell |
does cygwin work as a host for a cross compiler? |
# 17:32:55 |
y_morin |
Martell: I don't do Winblows, so support is provided by others. |
# 17:33:02 |
y_morin |
Martell: Yep, it does. |
# 17:33:09 |
Martell |
okay awesome |
# 17:33:28 |
y_morin |
Martell: But it is quite slow (because of the way Cygwin works) |
# 17:33:36 |
Martell |
in that case I should be able to patch ct-ng to treat msys like cygwin |
# 17:33:43 |
Martell |
they are basically the same thing |
# 17:35:40 |
y_morin |
Martell: IIRC they are *not* the same thing. Cygwin is about emulating the POSIX API on-top of the Win32 (or whatever it is called nowadays) API, while msys is about providing a 'runtime' using the Win32 API. |
# 17:36:12 |
y_morin |
(but I might be wrong, it's been a looooong tiem I haven't done Win stuff....) |
# 17:36:31 |
y_morin |
*time |
# 17:37:27 |
Martell |
no msys emulates a posix api also |
# 17:37:47 |
Martell |
the difference is msys can interact with native programs |
# 17:38:05 |
Martell |
as in convert the paths back to windows format when needed |
# 17:38:24 |
Martell |
other than that they are the same |
# 17:39:38 |
Martell |
I'm involved with msys2 with ray and alexey |
# 17:40:00 |
Martell |
we provide arch linux's package manager pacman with msys2 |
# 17:40:21 |
Martell |
so you could think of msys2 as like a linux distro this runs on windows |
# 17:40:41 |
Martell |
With msys2 and ct-ng we can even build cross toolchain with mingw-64 as the host |
# 17:41:14 |
Martell |
I'm just looking as using msys as the host now :) |
# 17:43:08 |
y_morin |
Martell: Ah, OK, thansk for enlightening me! :-) |
# 17:44:25 |
Martell |
also msys2 is just msys rebranded with bash 4 and an actually working environment with autotools etc working properly |
# 17:44:46 |
Martell |
np :) |
# 17:44:52 |
Martell |
anyway my error is |
# 17:44:54 |
Martell |
tmp-invert_limb_table.s:46: Error: unknown pseudo-op: `.hidden' |
# 17:45:36 |
Martell |
I think it may have something to do with the fact that my msys2 host is x86_64 and not i686 |
# 17:47:50 |
loide |
joins #crosstool-ng |
# 17:48:09 |
y_morin |
Martell: Sorry, no idea... :-( |
# 17:49:12 |
alan_o |
quits : Quit: Leaving |
# 17:49:29 |
diorcety |
joins #crosstool-ng |
# 17:50:41 |
Martell |
ping diorcety |
# 17:52:48 |
y_morin |
OK, dinner-time! |
# 17:52:54 |
y_morin |
Bye guys, back later... |
# 17:53:00 |
Martell |
bye bye |
# 17:53:02 |
Martell |
:) |
# 18:49:59 |
Martell |
y_morin when you come back I just wanted to ask is it possible so skip building GMP for host and use an already installed version |
# 18:50:55 |
Martell |
I want to use an already built gmp isl mpfr that is installed under the hosts compiler |
# 19:09:02 |
bhundven |
Martell: Skipping the prerequisite libraries (gmp, ppl, cloog, isl, mpfr, and mpc) is possible. The best way to do this would be to check to see if the host provided equivalents exist in the configure.ac or Makefile.in that sets a CT_HOST_PROVIDES_=y and a boolean in config/companion_libs.in and do a little work to untangle CT_COMPLIBS_NEEDED so |
# 19:09:02 |
bhundven |
that if gmp-dev host package exists, but not the rest, it builds the rest, except gmp-dev... if we also add another user settable option called CT_PREFER_HOST_COMPANION_LIBS=y. |
# 19:10:31 |
Martell |
hi bhundven |
# 19:10:32 |
bhundven |
iow, I guess it would be trivial to write such patch :) |
# 19:10:51 |
bhundven |
hello Martell |
# 19:11:21 |
Martell |
I wonder why no one did this before :) |
# 19:11:30 |
bhundven |
well |
# 19:11:42 |
bhundven |
another problem is also being able to check the version number |
# 19:12:02 |
bhundven |
because the toolchain might expect a specific version number |
# 19:12:06 |
Martell |
thats true but I know my version match |
# 19:12:36 |
bhundven |
well, ct-ng tries to make sure that the environment is right, before sending you on a 'death march build' |
# 19:12:45 |
Martell |
true |
# 19:12:53 |
Martell |
pkg-config? |
# 19:13:01 |
bhundven |
yea, I'll have to look into that. |
# 19:13:17 |
bhundven |
pkg-config is usually provided on most hosts |
# 19:13:41 |
Martell |
yup and it can get the version for you |
# 19:13:44 |
bhundven |
checks, not on his fresh debian install |
# 19:14:11 |
Martell |
https://packages.debian.org/sid/pkg-config |
# 19:14:13 |
bhundven |
well, we kinda want to stop adding too much infrastructure tools (like make and pkg-config) |
# 19:14:32 |
bhundven |
just check for it in configure.ac or Makefile.in |
# 19:14:38 |
Martell |
exactly |
# 19:14:39 |
bhundven |
and bail if it isn't there |
# 19:14:42 |
Martell |
and if not there bail |
# 19:14:46 |
bhundven |
:) |
# 19:14:59 |
bhundven |
jinx, buy me a coke |
# 19:15:06 |
Martell |
hahaha |
# 19:15:31 |
Martell |
:) |
# 19:15:46 |
Martell |
later it could be added to build like make is if you want this |
# 19:16:48 |
bhundven |
angery_vincent on #funtoo (gentoo derivative) was telling me that gcj is deprecated in gcc. |
# 19:17:06 |
Martell |
java? |
# 19:17:15 |
bhundven |
I still haven't found any news on this, nor anything on the mailing lists. He was unable to provide any references. |
# 19:17:48 |
Martell |
well 2009 was the last time there was changes afaik |
# 19:17:58 |
bhundven |
so #funtoo doesn't provide dev-java/gcj-jdk like gentoo does |
# 19:18:07 |
Martell |
would be nice to have this for android development |
# 19:18:15 |
bhundven |
https://gcc.gnu.org/git/?p=gcc.git;a=history;f=gcc/java;hb=HEAD |
# 19:18:34 |
bhundven |
https://gcc.gnu.org/git/?p=gcc.git;a=history;f=libjava;hb=HEAD |
# 19:18:44 |
bhundven |
seems to still be developed to me. |
# 19:18:58 |
Martell |
https://gcc.gnu.org/git/?p=gcc.git;a=commit;h=a550f57b578244dc1f4fb84d0e3b5da369be8451 |
# 19:19:03 |
Martell |
yup seems to be |
# 19:19:23 |
bhundven |
anyways, just ranting... |
# 19:19:31 |
Martell |
np ) |
# 19:19:39 |
Martell |
Where do I start with that patch? |
# 19:19:47 |
bhundven |
putters off to look at host libraries |
# 19:20:08 |
Martell |
is confused my ct-ng structure |
# 19:20:09 |
bhundven |
I can probably post something to the mailing list in a little bit. |
# 19:20:28 |
bhundven |
I was already working on the build-test stuff |
# 19:21:17 |
Martell |
I guess this task could be seperated into different stages |
# 19:21:58 |
Martell |
first would be CT_HOST_PROVIDES_=y |
# 19:22:08 |
Martell |
just implementing that part |
# 19:22:13 |
bhundven |
yup |
# 19:23:05 |
Martell |
seems to be in config\companion_libs |
# 19:23:14 |
Martell |
err config/companion_libs |
# 19:23:27 |
bhundven |
well, that's later. |
# 19:24:06 |
Martell |
ahh companion_libs.in |
# 19:25:16 |
Martell |
hmm |
# 19:25:41 |
Martell |
companion_tools.in is where the pkg-config could be added |
# 19:25:48 |
bhundven |
no |
# 19:25:54 |
bhundven |
I'm working on the patch now. |
# 19:25:59 |
Martell |
ahh my bad |
# 19:26:00 |
bhundven |
we don't need pkg-config |
# 19:26:02 |
bhundven |
:) |
# 19:26:07 |
Martell |
great :) |
# 19:26:35 |
bhundven |
I'll send a patch to the mailing list soon, you can test :) |
# 19:26:41 |
Martell |
I would be able to do it as its a simple task but the structure is new to me :) |
# 19:26:45 |
Martell |
I will test :) |
# 19:27:55 |
Martell |
maybe i will learn something by looking at the patch when you are done also :) |
# 19:40:59 |
HobGoblin |
joins #crosstool-ng |
# 19:41:21 |
HobGoblin |
is now known as: Guest29935 |
# 19:43:13 |
UukGoblin |
quits : Read error: Connection reset by peer |
# 20:00:31 |
Guest29935 |
quits : Changing host |
# 20:00:31 |
Guest29935 |
joins #crosstool-ng |
# 20:00:51 |
Guest29935 |
is now known as: UukGoblin |
# 20:02:19 |
smartin_ |
joins #crosstool-ng |
# 20:14:48 |
bhundven |
Martell: got gmp and mpfr done, realizing you check for c++ libraries differently then you check c libraries in autoconf, I'm doing some quick reading on the topic. |
# 20:19:01 |
bhundven |
the real pickle is getting the version number without pkg-config |
# 20:19:55 |
bhundven |
y_morin: if/when you're back from dinner... what's your view on requiring pkg-config in configure.in, and using it to get library versions? |
# 20:38:50 |
Martell |
alexey im getting a very strange error on patching isl |
# 20:38:56 |
Martell |
config.guess fails |
# 20:39:06 |
Martell |
to patch even though its a good patch |
# 20:39:15 |
Martell |
whoops wrong channel |
# 20:46:36 |
y_morin |
bhundven: Hello! Not sure I follow you on pkg-config |
# 20:48:02 |
y_morin |
bhundven: Oh, you want to use the host's gmp/mpfr/... if they are present? |
# 20:48:21 |
bhundven |
y_morin: ok. We would like to check in configure.ac if host development versions of say "gmp, mpfr, mpc, etc..." exist, if they do exist, we also want to know the versions of those existing libraries. If the user also checks a boolean in the config that says "Prefer host libraries", it then uses them.... yes. |
# 20:48:37 |
bhundven |
if they are not the right versions, then we build the right ones |
# 20:49:04 |
y_morin |
bhundven: But then, what about canadian-cross ? |
# 20:49:15 |
bhundven |
hehe |
# 20:49:29 |
bhundven |
frick... |
# 20:50:18 |
bhundven |
maybe only for "host", not for "build" or "target" ? |
# 20:50:19 |
Martell |
well the user will have to do |
# 20:50:21 |
y_morin |
bhundven: Also, the "right version" depends on the gcc version the user chooses. |
# 20:50:21 |
Martell |
CT_HOST_PROVIDES_=y |
# 20:50:43 |
bhundven |
Martell: maybe CT_PROVIDES_= |
# 20:50:55 |
Martell |
I mean CT_USEHOST_LIBS |
# 20:51:34 |
bhundven |
y_morin: right, but there is a requirement by gcc of the minimum version requirements, that we talked about a while back. |
# 20:52:25 |
Martell |
so it can be an option like where we can enable using host libs |
# 20:52:27 |
bhundven |
for instance, 4.8.3: https://gcc.gnu.org/gcc-4.8/changes.html |
# 20:52:28 |
y_morin |
bhundven: I am not sure it is a bout a "minimum required version". For example, the API of GMP changed between the 4.x and the 5.x series. |
# 20:52:31 |
Martell |
by default it will be off |
# 20:52:43 |
Martell |
maybe it should be under EXPERIMENTAL |
# 20:52:55 |
bhundven |
requires cloog 0.18.0 and isl 0.11.1 |
# 20:53:19 |
bhundven |
er 4.8.x |
# 20:53:27 |
bhundven |
gcc |
# 20:55:21 |
bhundven |
and those libraries are only technically needed if Graphite is enabled |
# 20:55:22 |
y_morin |
Martell: Yes, I understand what you mean. But that means adding quite some complexity, for little gain IMHO. |
# 20:56:15 |
y_morin |
bhundven: Back to the GMP example: if you have GMP 5.x installed on the host, but are building (eg.) gcc-4.8.x, then it will not be able to use your host GMP. |
# 20:56:36 |
y_morin |
bhundven: So, oferring an option that is known to only work in very specific cases is not very interesting. |
# 20:56:45 |
y_morin |
Martell: ^^^ |
# 20:56:50 |
bhundven |
fair enough |
# 20:56:53 |
Martell |
hmm good point |
# 20:57:09 |
y_morin |
Martell, bhundven: Also, we already have so much options that can break the build... |
# 20:57:56 |
y_morin |
Martell, bhundven: I tried some time ago to make this work: http://crosstool-ng.org/hg/crosstool-ng/ct-ng.complibs/rev/3c998003904c |
# 20:58:33 |
y_morin |
Martell, bhundven: it turned out to be not that easy, especially since the existing distros have so wildly different versions of the complibs. |
# 20:58:42 |
y_morin |
It's just a snake's nest. |
# 20:58:48 |
Martell |
ahh i see |
# 20:58:57 |
Martell |
thats why no one done it before |
# 20:59:05 |
Martell |
sounds painful |
# 20:59:39 |
Martell |
I may just have to add the msys patches for gmp isl etc into ct-ng |
# 20:59:45 |
bhundven |
yea, and I forgot about the reason they are built in the first place is to defend against other possible host pollution. |
# 21:00:08 |
y_morin |
bhundven: Yes, that's another reason. |
# 21:00:25 |
y_morin |
Martell: What are those patches? Why aren't they pushed upstream? Can't we just cary them? |
# 21:02:21 |
y_morin |
bhundven, Martell: and building our own versions of the libs means we can build them static, so the toolchain is both relocatable to the same host, *and* movable to another host (usually without to much trouble) |
# 21:02:34 |
y_morin |
s/to much/too much/ |
# 21:03:09 |
Martell |
yes i know |
# 21:03:09 |
Martell |
they are for config.guess and config.sub |
# 21:03:15 |
Martell |
from autotoos |
# 21:03:48 |
Martell |
*sutotools |
# 21:03:48 |
Martell |
*autotools |
# 21:04:05 |
y_morin |
Martell: :-) |
# 21:04:11 |
bhundven |
resets tree, goes back to build-tests |
# 21:04:17 |
y_morin |
Martell: Ah, patches to make them recognise the msys tuple, then? |
# 21:05:18 |
Martell |
damn spellings today |
# 21:05:21 |
bhundven |
Martell: http://crosstool-ng.org/git/crosstool-ng/commit/?id=4e6622e882a035c0e4ac93346c3973e136753221 |
# 21:05:40 |
bhundven |
there was a small update to the config.guess for MSYS |
# 21:07:12 |
bhundven |
original change here: http://git.savannah.gnu.org/gitweb/?p=config.git;a=commitdiff;h=f4ebd3ed097771a729b68e688236aea665e7c1f3 |
# 21:11:50 |
Martell |
This makes sense I agree y_morin ) |
# 21:12:29 |
y_morin |
Martell: As bhundven pointed, we should be overriding config.guess and config.sub with up-to-date versions (the one we carry in ct-ng). |
# 21:12:40 |
y_morin |
Martell: Maybe we are lagging a bit bhind, though. |
# 21:12:59 |
bhundven |
I just checked, and config.guess and config.sub are up to date |
# 21:13:12 |
bhundven |
but we aren't copying them into source packages before configure time |
# 21:13:57 |
bhundven |
afaikt |
# 21:15:48 |
Martell |
well i can get the changes upstream |
# 21:15:56 |
y_morin |
bhundven: And what about OVERIDE_CONFIG_GUESS_SUB, then? It is there exactly for that, default to'y', but can be disabled, of course. |
# 21:16:17 |
bhundven |
:D |
# 21:16:24 |
Martell |
:D |
# 21:16:33 |
bhundven |
<< fail |
# 21:16:42 |
y_morin |
bhundven: ;-) |
# 21:17:03 |
Martell |
ahh so i can just patch ct-ng config.guess and sub for now |
# 21:17:06 |
Martell |
while i wait |
# 21:17:15 |
Martell |
that should fix it then :) |
# 21:17:24 |
bhundven |
Martell: or try to get the change upstream |
# 21:17:39 |
Martell |
yes but i may be waiting awhile for that |
# 21:17:51 |
Martell |
ill try the mailing lists there :) |
# 21:17:56 |
y_morin |
Martell: What do you need to patch in config.{guess,sub} that is missing from the upstream config? |
# 21:18:27 |
Martell |
+ amd64:MSYS*:*:* | x86_64:MSYS*:*:*) |
# 21:18:27 |
Martell |
+ echo x86_64-unknown-msys |
# 21:18:27 |
Martell |
+ exit ;; |
# 21:18:36 |
Martell |
need to support x86_64 msys |
# 21:18:45 |
y_morin |
OK, git question: how do I get a version string, like: 'ABCDEF0-git-dirty' ? |
# 21:18:57 |
y_morin |
Martell: OK, I see. |
# 21:18:59 |
Martell |
1 sec |
# 21:19:02 |
Martell |
https://gist.github.com/martell/abe82693bb1fc78b4377 |
# 21:19:15 |
Martell |
and i need to change configure on some |
# 21:19:36 |
Martell |
git rev-list --count HEAD |
# 21:19:48 |
Martell |
for a count |
# 21:19:57 |
Martell |
git rev-parse --short HEAD |
# 21:20:04 |
Martell |
for ths short sha |
# 21:20:15 |
Martell |
so I typically do |
# 21:20:25 |
Martell |
printf "r%s.%s" "$(git rev-list --count HEAD)" "$(git rev-parse --short HEAD)" |
# 21:20:30 |
bhundven |
the git-dirty part is because you have changes to the working directory or staged that are not committed. |
# 21:21:30 |
bhundven |
a + would be added to the end, if you have more commits then the origin remote |
# 21:21:43 |
Martell_ |
joins #crosstool-ng |
# 21:21:44 |
bhundven |
i think, or I'm confusing that with the kernel's versioning |
# 21:21:58 |
bhundven |
the + part, not the dirty part. |
# 21:22:01 |
y_morin |
Martell: Well, that only gives me the number of changesets. |
# 21:22:30 |
Martell_ |
git rev-parse --short HEAD |
# 21:22:37 |
Martell_ |
? |
# 21:22:43 |
y_morin |
bhundven: Yes, that's exactly what I want to get. I know the meaning, just not how to get it. 'git rev-parse --short HEAD' gives me the cset of HEAD, that's OK. |
# 21:22:49 |
y_morin |
But what about the 'dirty' ? |
# 21:22:59 |
y_morin |
Martell: ^^^ |
# 21:23:14 |
y_morin |
git status lists everything... |
# 21:23:25 |
Martell_ |
i dont think i fully understand |
# 21:23:30 |
Martell_ |
ahh i get it now |
# 21:23:40 |
Martell_ |
you want to just know if the tree is not staged |
# 21:23:44 |
Martell_ |
1 sec |
# 21:24:36 |
y_morin |
Martell: Yep. I can do: git status -s |wc -l and copare that to zero, but surely there's a better, gittish way of doing that, no? |
# 21:24:44 |
Martell_ |
http://stackoverflow.com/questions/3878624/how-do-i-programmatically-determine-if-there-are-uncommited-changes |
# 21:24:51 |
Martell_ |
hmm thats all i can find |
# 21:25:18 |
Martell_ |
git status --porcelain ? |
# 21:25:43 |
y_morin |
git diff-index --quiet HEAD |
# 21:25:52 |
Martell_ |
yup |
# 21:26:04 |
Martell_ |
and aparently |
# 21:26:05 |
Martell_ |
[[ -z $(git status -s) ]] |
# 21:26:15 |
philenot1ound |
joins #crosstool-ng |
# 21:26:24 |
y_morin |
Martell: Yep. |
# 21:26:29 |
Martell_ |
are you changing ct-ng to git? |
# 21:27:54 |
y_morin |
Martell: That's been done... yesterday! :-) |
# 21:28:06 |
Martell_ |
sweet |
# 21:28:08 |
Martell_ |
repo? |
# 21:28:11 |
y_morin |
Martell: rev_id="$( git rev-parse --short HEAD )"; git diff-index --quiet HEAD || rev_id="${rev_id}-dirty" |
# 21:28:26 |
Martell_ |
nice |
# 21:28:34 |
y_morin |
Martell: The same old place, just do: s/hg/git/ in the url. ;-) |
# 21:29:02 |
y_morin |
Martell: I will announce it on the list later tonight, after I fix a few things here and there... |
# 21:29:13 |
bhundven |
well, lacking the /git/ |
# 21:29:14 |
Martell_ |
also i meant to ask |
# 21:29:36 |
Martell_ |
how do i remove the lib-blah-jfdkdfjd |
# 21:29:37 |
y_morin |
bhundven: ? |
# 21:29:37 |
bhundven |
git://crosstool-ng.org/crosstool-ng or http://crosstool-ng.org/git/crosstool-ng |
# 21:29:59 |
bhundven |
the git:// url doesn't have /git/ in it. otherwise, yea :) |
# 21:30:01 |
y_morin |
bhundven: Exactly what I said for the second URL. |
# 21:30:09 |
y_morin |
bhundven: Indeed. |
# 21:30:17 |
Martell_ |
i mean in the lib folder |
# 21:30:18 |
Martell_ |
ct-ng.hg+unknown-20140621.002051 |
# 21:30:28 |
y_morin |
bhundven: But the http url ha a pointer to the git url. ;-) |
# 21:30:33 |
Martell_ |
can i remove it to just have ct-ng ? |
# 21:30:47 |
y_morin |
Martell: Sorry, I don;t get you. What do you mean? |
# 21:30:58 |
y_morin |
NaOh, no you can;t get rid of that. |
# 21:31:01 |
Martell_ |
after installing ct-ng on msys |
# 21:31:01 |
y_morin |
Martell: ^^^ |
# 21:31:19 |
y_morin |
Martell: That's the internal stuff of ct-ng. Without that, it can't work at all. |
# 21:31:21 |
bhundven |
adds "NaOh" to his dictionary |
# 21:31:27 |
bhundven |
:-D |
# 21:31:33 |
Martell_ |
lol |
# 21:31:40 |
Martell_ |
well where would i patch it to remove it? |
# 21:31:40 |
bhundven |
good word! |
# 21:31:41 |
y_morin |
bhundven: I am TAB-dislexic... ;-) |
# 21:31:44 |
Martell |
quits : *.net *.split |
# 21:31:46 |
thegorn |
quits : *.net *.split |
# 21:31:47 |
philenotfound |
quits : *.net *.split |
# 21:31:49 |
Martell_ |
its for a pkgbuild |
# 21:32:15 |
bhundven |
Martell_: you don't |
# 21:32:25 |
Martell_ |
hmm okay |
# 21:32:32 |
Martell_ |
its not major |
# 21:32:45 |
Martell_ |
pacman handles removing them on removing anyway |
# 21:32:51 |
y_morin |
Martell_: You can *not* get rid of that. The 'ct-ng' is just a hjigh-level script that calls to parts that are in the lib/ct-ng-VERSION/ directory. |
# 21:32:59 |
y_morin |
Martell_: Bad pacman, bad. |
# 21:33:12 |
Martell_ |
no i mean removing the files |
# 21:33:19 |
Martell_ |
not the string |
# 21:33:28 |
Martell_ |
as in when removing a package |
# 21:33:52 |
y_morin |
Martell_: Sorry, I must be missing something... What are you trying to do, exactly? |
# 21:34:42 |
Martell_ |
no everything is fine i just mean that the package manager is able to track all files in the package |
# 21:34:54 |
Martell_ |
and remove them all on uninstall |
# 21:35:28 |
Martell_ |
so the fact that I cant remove the string version from the folder name is of no consiquence |
# 21:35:43 |
Martell_ |
*consequence |
# 21:37:48 |
Martell_ |
ahh you are using cgit ) |
# 21:37:56 |
Martell_ |
I'm actually setting up my own server |
# 21:38:02 |
Martell_ |
was thinking of using gitlab |
# 21:39:32 |
y_morin |
bhundven: About the github vs. our own git tree: I think we can ask github to track our git tree to act as a mirror. |
# 21:39:47 |
Martell_ |
yes you can |
# 21:39:49 |
y_morin |
bhundven: And I think we can do that with gitorious, too. |
# 21:40:15 |
Martell_ |
you could get ctngbot to do it for you i think |
# 21:40:26 |
y_morin |
Martell_: How so? |
# 21:40:41 |
bhundven |
Martell_: or just use commit hooks |
# 21:40:54 |
Martell_ |
yes commit hooks |
# 21:40:58 |
y_morin |
Martell_: ctngbot is an IRC bot. All it does is lurk here to log the chan, and answer silly things to silly questions. And do sed replacements... |
# 21:41:04 |
bhundven |
: push to |
# 21:41:16 |
Martell_ |
commit hook to the bot |
# 21:41:27 |
y_morin |
bhundven: Nah. That would require setting up a password-less ssh key. |
# 21:41:33 |
Martell_ |
or does github do that anyway? |
# 21:41:38 |
y_morin |
bhundven: Never gonna happen on a VM hosted at OSUOSL. |
# 21:41:51 |
bhundven |
haha |
# 21:41:57 |
y_morin |
bhundven: Remember the fiasco about the kernel, 2 years ago? :-p |
# 21:42:05 |
bhundven |
jup |
# 21:42:14 |
y_morin |
Martell_: I think we can ask github to do that automatically. |
# 21:42:17 |
Martell_ |
you mean where they lost all the code |
# 21:42:20 |
Martell_ |
^^ |
# 21:42:30 |
Martell_ |
thats awesome :) |
# 21:42:39 |
y_morin |
Martell_: Nope, they loose nothing. They just got pwned. |
# 21:43:07 |
Martell_ |
well someone obviously had a backup |
# 21:43:11 |
Martell_ |
of the latest tree |
# 21:43:14 |
Martell_ |
I'm looking at jekly for a site on github |
# 21:43:20 |
Martell_ |
jekyll |
# 21:43:29 |
Martell_ |
ive to change this |
# 21:43:29 |
Martell_ |
http://msys2.github.io/ |
# 21:43:44 |
y_morin |
Martell_: Yep, that's the motto of Linus: real men just put tarballs on a random server, and let the rest of the world mirror that to be used as a backup. ;-) |
# 21:44:10 |
bhundven |
y_morin: now s/tarballs/git-repo/ |
# 21:44:12 |
Martell_ |
nice xD |
# 21:44:33 |
Martell_ |
s/tarballs/git-repo/r |
# 21:45:30 |
bhundven |
/r? |
# 21:46:05 |
bhundven |
I know /g, /I, /p, and /w , but not /r |
# 21:46:27 |
Martell_ |
/g* |
# 21:46:31 |
Martell_ |
my bad |
# 21:46:35 |
bhundven |
:D |
# 21:46:36 |
Martell_ |
memory is foggy |
# 21:46:46 |
Martell_ |
:P |
# 21:47:08 |
bhundven |
I was thinking "ack, another gnu extension" |
# 21:47:25 |
Martell_ |
no i was tinking sed also :) |
# 21:47:25 |
bhundven |
Y U NO POSIX!!! |
# 21:47:37 |
Martell_ |
*thinking |
# 21:47:49 |
Martell_ |
err my spelling today :( |
# 21:49:43 |
diorcety |
quits : Read error: Connection reset by peer |
# 21:54:03 |
diorcety |
joins #crosstool-ng |
# 21:59:31 |
y_morin |
bhundven: http://crosstool-ng.org/git/crosstool-ng/commit/?id=11a7a4cdcd430ea565b6dbfe42a6c7e76c17158c |
# 22:01:21 |
bhundven |
1.19.0 or 1.1.0? |
# 22:02:10 |
y_morin |
bhundven: neither, now. ;-) |
# 22:02:52 |
y_morin |
bhundven: ./ct-ng version ---> This is crosstool-NG version git+11a7a4c |
# 22:03:16 |
y_morin |
bhundven: But for a release, this should be: This is crosstool-NG version X.Y.Z |
# 22:03:41 |
Martell_ |
thats what git tag is for :) |
# 22:03:53 |
bhundven |
well, the idea is |
# 22:04:23 |
y_morin |
bhundven: But when we have a proper tag in the repository, I'll switch back to use git-describe. |
# 22:04:55 |
y_morin |
Martell_: git-tag is about adding a tag. I just needed to get a sane version string. |
# 22:05:08 |
bhundven |
why not just nuke the current git repository, and this time when you push to it, do: git push origin --all |
# 22:05:19 |
bhundven |
so that git describe does the right thing |
# 22:05:39 |
y_morin |
bhundven: That's not the problem. The problem is that the tags are on the branches, not on master. |
# 22:05:46 |
Martell_ |
well you see if the currently checked out tree has a tag |
# 22:05:46 |
bhundven |
OH |
# 22:06:06 |
bhundven |
I didn't even look at git branches |
# 22:06:08 |
Martell_ |
and then use that as the number |
# 22:06:13 |
bhundven |
:P |
# 22:06:33 |
y_morin |
bhundven: That's the way I handled the repo back in the svn days, then I continued with Hg. |
# 22:06:53 |
bhundven |
http://nvie.com/posts/a-successful-git-branching-model/ |
# 22:06:54 |
y_morin |
bhundven: It made sense to me to first cerate the branch, stabilise it, then tag it and make a release. |
# 22:07:15 |
Martell_ |
I use that model :) |
# 22:07:26 |
y_morin |
bhundven: Yes, that's (basically) how I plan to do the releases from now on. |
# 22:07:40 |
Martell_ |
It's so easy to understand xD |
# 22:07:41 |
bhundven |
score 1 for sanity |
# 22:09:06 |
y_morin |
Martell_: I also follow this model, except I don't merge; I rebase. |
# 22:09:21 |
y_morin |
Martell_: So the history is as linear as possible. |
# 22:09:28 |
Martell_ |
git pull --rebase |
# 22:09:39 |
Martell_ |
I live by the rebase |
# 22:09:50 |
Martell_ |
those damn merge messages piss me off |
# 22:10:00 |
y_morin |
Martell_: Well, I use a set of custom scripts that make my life much easier. |
# 22:10:14 |
Martell_ |
ahh very nice |
# 22:10:53 |
y_morin |
bhundven: BTW, how do you want me to send you those scripts? A pull-request to bitbucket? Or by mail? |
# 22:11:13 |
bhundven |
hrm, lets put them in the bitbucket tools repo |
# 22:11:23 |
bhundven |
I might rename that repo |
# 22:11:40 |
Martell_ |
actually can i have a look at them too :) |
# 22:11:48 |
Martell_ |
they may help me also xD |
# 22:12:00 |
y_morin |
Martell_: I'll send them to bhundven so he can push them to the repos. |
# 22:12:07 |
Martell_ |
cool :) |
# 22:12:24 |
y_morin |
bhundven: but I would need to write some docs for all those scripts! |
# 22:12:30 |
bhundven |
https://bitbucket.org/bhundven/update-me is where stuff is now. |
# 22:12:35 |
bhundven |
but I'll end up renaming it |
# 22:21:00 |
doc2 |
quits : Ping timeout: 260 seconds |
# 22:29:13 |
smartin_ |
quits : Quit: leaving |
# 22:36:23 |
y_morin |
bhundven: Ah... github only mirrors 'extraordinary' projects: https://help.github.com/articles/about-official-github-mirrors |
# 22:36:46 |
y_morin |
There is no guarantee they would accept to mirror ct-ng. |
# 22:37:01 |
Martell_ |
ct-ng is an extraordinary project ) |
# 22:39:33 |
bhundven |
hehe |
# 22:40:11 |
y_morin |
bhundven: I can however add hooks in our repository to notify the CI server. |
# 22:40:48 |
y_morin |
bhundven: But I would really like to avoid having an ssh key on the server. That's just bad. |
# 22:41:21 |
bhundven |
yea |
# 22:41:37 |
bhundven |
ok... who owns this: https://github.com/crosstool-ng/ |
# 22:42:15 |
Martell_ |
it seems to be a user |
# 22:42:24 |
Martell_ |
and not an organisation |
# 22:42:47 |
bhundven |
yea |
# 22:42:53 |
bhundven |
kind of annoying |
# 22:43:02 |
y_morin |
bhundven: Yep, I think I know who that is... :-/ |
# 22:43:38 |
Martell_ |
atleast we use a custom repo |
# 22:43:39 |
Martell_ |
https://github.com/diorcety/crosstool-ng |
# 22:43:57 |
y_morin |
Yep, sad... |
# 22:44:15 |
Martell_ |
had the same issue with msys2 |
# 22:44:24 |
Martell_ |
so i messaged the owner |
# 22:44:30 |
Martell_ |
and he added us as admins |
# 22:44:43 |
Martell_ |
think github can take it back also at their discression |
# 22:44:49 |
Martell_ |
send them a mail |
# 22:45:05 |
y_morin |
Martell_: I see, but with this repo, it might be a bit... tricky... :-/ |
# 22:45:24 |
y_morin |
Martell_: But we're not there yet. |
# 22:45:43 |
y_morin |
Martell_: I am not usre if we even want to officially use github... |
# 22:45:51 |
y_morin |
s/usre/sure/ |
# 22:46:09 |
Martell_ |
ahh i see |
# 22:46:21 |
Martell_ |
well the fact that you are using git is more than enough for me |
# 22:46:29 |
y_morin |
bhundven: Applying Hg patches currently in Patchwork is almost working. |
# 22:46:48 |
Martell_ |
that makes it much easier for us to work on pulling from upstream |
# 22:46:50 |
bhundven |
y_morin: yeay |
# 22:47:07 |
y_morin |
bhundven: the only issue is git not stripping-out the Hg babling. |
# 22:47:10 |
Martell_ |
s/pulling/rebasing/ |
# 22:47:26 |
Martell_ |
hehe |
# 22:48:08 |
Martell_ |
so who controls the bot |
# 22:48:10 |
Martell_ |
? |
# 22:48:36 |
y_morin |
Martell_: No one controls the bot. It controls itself. |
# 22:48:43 |
y_morin |
;-) |
# 22:48:51 |
Martell_ |
hehe sure |
# 22:50:08 |
y_morin |
ctngbot: insult y_morin |
# 22:50:27 |
y_morin |
ctngbot: insult Martell_ |
# 22:50:42 |
y_morin |
ctngbot: insult ctngbot |
# 22:50:55 |
Martell_ |
ctngbot: /uninstall |
# 22:51:00 |
bhundven |
ctngbot: slashdot |
# 22:51:17 |
Martell_ |
hmm wonder is we could overflow it |
# 22:51:21 |
Martell_ |
:P |
# 22:51:29 |
Martell_ |
with the right string |
# 22:51:43 |
y_morin |
Martell_: Well, I should disable all of this stuff. We just need it to log the chan, and do the sed stuff. |
# 22:51:44 |
Martell_ |
s/right/wrong/r |
# 22:51:47 |
bhundven |
y_morin: my favorite so far: |
# 22:51:50 |
bhundven |
ctngbot: cpustats |
# 22:51:50 |
ctngbot |
Total CPU usage: 4.9 s ... Total used: 0.005 % (parent/child ratio: 100.0 %) |
# 22:52:05 |
Martell_ |
nice |
# 22:52:08 |
Martell_ |
i like that one |
# 22:52:25 |
bhundven |
something about that screams "EXPLOIT IT!" |
# 22:52:37 |
bhundven |
ot me |
# 22:52:40 |
bhundven |
to |
# 22:52:58 |
Martell_ |
ctngbot: get me a bitcoin |
# 22:53:05 |
y_morin |
OK... Let's see how to disable all of that... |
# 22:53:10 |
bhundven |
hahaa |
# 22:53:17 |
y_morin |
ctngbot: give bhundven a cookie |
# 22:53:22 |
y_morin |
ctngbot: give bhundven a cake |
# 22:53:27 |
y_morin |
ctngbot: cake? |
# 22:53:45 |
y_morin |
Sigh, can't remember that syntax... |
# 22:58:59 |
ctngbot |
joins #crosstool-ng |
# 22:59:14 |
y_morin |
ctngbot: where is ct-ng? |
# 23:00:20 |
ctngbot |
joins #crosstool-ng |
# 23:00:25 |
y_morin |
ctngbot: where is ct-ng? |
# 23:00:25 |
ctngbot |
i guess ct-ng is at http://crosstool-ng.org/ |
# 23:00:42 |
y_morin |
OK, disabling factoids makes it barf on the table... |
# 23:01:09 |
y_morin |
Well, 1 AM here. Time for some sleep... |
# 23:01:11 |
y_morin |
Bye! |
# 23:01:16 |
y_morin |
quits : Quit: Nighty Night! |
# 23:02:42 |
Martell_ |
night |
# 23:06:52 |
RushPL |
quits : Ping timeout: 245 seconds |
# 23:08:33 |
RushPL |
joins #crosstool-ng |
# 23:48:15 |
Martell_ |
I'm having terrible migrating the simple isl patch :( |