Parrot 3.3.0 released | parrot.org | Log: irclog.perlgeek.de/parrot/today” | Accepted GSoC Students announced! | GSoC student information emails coming out soon
Set by moderator on 26 April 2011.
kid51 OSU OSL understands that these problems persist. They will try to look at this later tonight, but right now it's Friday 5:00 PM at OSU, so we can't get this solved immediately. 00:04
00:17 whiteknight joined
kid51 You can now request a new Trac password that will enable you to log in, but the underlying problem persists, so don't count on that password working indefinitely. 00:18
cotto_work kid51: thanks 00:19
dukeleto: ping
00:20 benabik joined
dukeleto cotto_work: pong 00:23
$ hexdump -C t/m0/hello.m0b
00000000 fe 4d 30 42 0d 0a 1a 0a 00 04 08 04 04 00 00 00 |.M0B............|
cotto_work dukeleto: what do you think about adding labels to M0? Semantically meaningful labels would be nice, but grammatically insignificant labels would be fine.
dukeleto++
awesomeful
dukeleto cotto_work: i know nazzing, lebowski 00:24
cotto_work I'm writing tests for goto and it gets unreadable really quickly.
dukeleto cotto_work: you mean metadata labels?
cotto_work foo: add_i I0, I1, I2
It adds some complexity to the assembler though. 00:25
It'd also mean that the bytecode segment would need to be parsed before the variables table, or that parsing the bc segment would add entries to the variables segment. 00:26
Though to generate correct m0b you need to parse the whole file before emitting anything anyway. 00:27
It'd also mean having a basic register allocator to avoid stepping on any live registers. 00:32
dalek rrot/m0-prototype: 3405cae | dukeleto++ | / (2 files):
Add some POD
 00:33
rrot/m0-prototype: a930740 | dukeleto++ | / (2 files):
Fix some bugs, add some tests, and generate a proper M0 bytecode header

A bug was fixed where invalid bytecode files where created and a function to generate a M0 binary bytecode header was borrowed from one of cotto++'s M0 interpreter test files.
cotto_work It generates headers? 00:36
Ship it!
00:39 jsut_ joined
dukeleto cotto_work: it generates partial m0 bytecode which includes a header and then some malformed m0 bytecode :) 00:41
cotto_work dukeleto: great. I've been jonesing for some malformed bytecode. 00:42
dalek rrot/m0-prototype: 25c8a51 | cotto++ | t/m0/m0_interp.t:
add a test for goto_if_eq

Note that labels aren't in the M0 spec so this test won't pass, even once the assembler works. Once the assembler and interp are far enough along to fail this test, we can decide where to go with this.		 00:44
00:44 jsut left
cotto_work and there's some malformed M0 for you.
00:57 bubaflub joined
bubaflub ~ 00:58
dalek nxed: r976 | NotFound++ | trunk/winxedst1.winxed:
several minor simplifications and redundance avoiding
 01:06
rrot/m0-prototype: 074e289 | dukeleto++ | / (3 files):
Fix bug in regex to parse textual bytecode lines, add a test
 01:16
01:20 soh_cah_toa joined 01:21 davidfetter left 01:25 kid51 left
cotto ~~ 01:40
whiteknight )) 01:41
benabik syntax error: -:0:1 - unmatched ")" 01:43
01:49 rurban_ joined 01:51 rurban left, rurban_ is now known as rurban 01:58 whiteknight left 02:11 dolmen left 02:18 krunen left, krunen joined 02:37 TiMBuS left 02:50 lateau joined 02:53 lateau left
cotto I need to buy a beer for whoever came up with the idea of having the Ubuntu installer ask for information while it formats the hdd. 03:02
It's also nice that the 7th CD I burned didn't have any errors.
03:02 jrt4__ joined
dukeleto cotto: is a chunk without a name invalid ? 03:51
cotto dukeleto, I think that should be the case 04:16
new machine takes 1:50 to build rakudo with -j 10 04:18
I like where this is going
spectest_regression just took less than 7 minutes. Best. Machine. Evar. 04:45
04:55 bubaflub left 05:14 soh_cah_toa left 05:29 ilbot2 joined
moderator Parrot 3.3.0 released | parrot.org | Log: irclog.perlgeek.de/parrot/today” | Accepted GSoC Students announced! | GSoC student information emails coming out soon
05:38 ilbot2 joined
moderator Parrot 3.3.0 released | parrot.org | Log: irclog.perlgeek.de/parrot/today” | Accepted GSoC Students announced! | GSoC student information emails coming out soon
05:40 soh_cah_toa joined 05:42 soh_cah_toa left 05:44 mikehh_ is now known as mikehh, woosley joined, bubaflub joined
mikehh opbots, names 05:44
05:45 woosley left 05:46 theory left 05:51 bubaflub left
dalek rrot/m0-prototype: fa4122a | dukeleto++ | src/m0/m0_assembler.pl:
Steal directory bytecode generation code from m0 interp tests
 06:01
rrot/m0-prototype: 6f3183e | dukeleto++ | / (5 files):
Update M0 test data to include chunk names
cotto dukeleto, I've made a note to myself to update that code. It'll need to be changed unless we want to say that M0 always uses 4-byte ints. 06:04
dukeleto, what's broken about the TODO'd code? 06:06
06:17 fperrad joined 06:22 particle left 06:34 benabik left 06:39 SHODAN joined 06:52 birdwindupbird joined 07:56 jrt4__ left
bacek ~~ 07:56
08:10 Kulag left, Kulag joined, nopaste left 08:12 jsut_ left, TonyC_ left, TonyC joined, jsut joined 08:13 nopaste joined, sjn joined 08:21 dodathome joined 08:29 AzureStone left 08:30 AzureStone joined 08:31 perlite_ joined 08:33 nopaste left, atrodo left, tewk left, tewk joined 08:34 ingy left, dukeleto left, ingy joined, perlite left, perlite_ is now known as perlite, atrodo joined, aloha left, bacek_at_work left 08:35 bacek_at_work joined 08:36 aloha joined 08:37 nopaste joined 09:03 frodwith left, tewk left, tewk joined 09:04 frodwith joined 09:13 contingencyplan left 09:14 contingencyplan joined 09:41 M_o_C joined
dalek rrot: 6f0cfa8 | bacek++ | src/gc/gc_gms.c:
Improve GMS performance by doing quick and cheap checks first during C stack scan.
 09:43
rrot: bd0a529 | bacek++ | src/pmc/capture.pmc:
Explicitely WB dest in Capture.clone. GC can be triggered between allocating of dest and cloning of guts.
09:50 rurban_ joined 09:51 rurban left 09:52 rurban_ is now known as rurban 10:10 contingencyplan left 10:59 Themeruta is now known as NotFound, whiteknight joined 11:04 lateau joined
whiteknight good morning, #parrot 11:06
msg soh_cah_toa for GSOC yu *must* have a blog on parrot.org. That isn't optional. You may also have a second blog somewhere else to share posts, but at least one copy of your posts should be on parrot.org. I can help you set up a blog on github too, if you want 11:07
aloha OK. I'll deliver the message.
12:08 Patterner left 12:16 woosley joined
dalek nxed: r977 | NotFound++ | trunk/winxedst1.winxed:
some more minor simplifications
 12:19
12:30 lateau1 joined
bacek seen chromatic 12:33
aloha chromatic was last seen in #parrot 84 days 18 hours ago saying "All of the core tests did pass for me, FWIW.".
bacek sigh...
12:35 lateau left 12:36 ambs joined
dalek rrot: 2953196 | bacek++ | src/pmc/callcontext.pmc:
Temporary break CallContext.clone to bring more attention.

Sorry, I'm kind of tired now to properly fix it after 3 hours of debugging session hunting this bug. But at least I nailed it down.		 12:40
moderator Parrot 3.3.0 released | parrot.org | Log: irclog.perlgeek.de/parrot/today” | Accepted GSoC Students announced! | GSoC student information emails coming out soon | Please fix master branch after github.com/parrot/parrot/commit/2953196d6c 12:42
moderator Parrot 3.3.0 released | parrot.org | Log: irclog.perlgeek.de/parrot/today | Accepted GSoC Students announced! | GSoC student information emails coming out soon | Please fix master branch after github.com/parrot/parrot/commit/2953196d6c 12:43
moderator Parrot 3.3.0 released | parrot.org | Log: irclog.perlgeek.de/parrot/today | Please fix master branch after github.com/parrot/parrot/commit/2953196d6c
12:46 ambs left 12:48 lateau joined
whiteknight bacek: I'll start looking at it soon 12:49
bacek whiteknight, thanks. Just implement "deep clone" of hash values. Like 10 lines of code, but I can't do it right now. 12:50
whiteknight okay, sounds easy enough. Go get some rest 12:51
bacek good idea
g'night
12:54 lateau1 left 13:23 varta_ left
dalek nxed: r978 | NotFound++ | trunk/examples/Mysql.winxed:
fix, improve and update coding style example Mysql
 13:43
p: 2656070 | masak++ | src/ModuleLoader.pm:
[ModuleLoader] strip away Unixy './' in module path

This prevented the lookup from working on Linux. Works now.		 13:46
14:03 ambs joined 14:15 lucian joined 14:17 M_o_C left 14:20 M_o_C joined
dalek p: dcd196e | jonathan++ | src/NQP/Grammar.pm:
Toss unused variable.
 14:59
15:12 whiteknight left 15:14 lucian_ joined 15:15 lateau1 joined 15:17 dukeleto joined
dalek nxed: r979 | NotFound++ | trunk/winxedst1.winxed:
coding style fixes and a few simplifications
 15:17
dukeleto cotto: i copied it directly from your interp tests, but my chunks do not have the same structure as your chunks 15:18
15:18 lucian left 15:19 lateau left
dukeleto cotto: is a chunk with no name invalid? 15:20
cotto: is the empty string a valid name?
cotto: are only double quotes allowed?
cotto: what happens when two chunks have the same name?
15:27 spinclad_ is now known as spinclad 15:28 M_o_C left
dalek rrot/m0-prototype: b0d177e | dukeleto++ | t/m0/m0_assembler.t:
Add a test for the size of the hello.m0b file
 15:29
cotto so man questions 15:32
many
15:32 bubaflub joined
cotto dukeleto, my current thinking is that chunk names are the primary mechanism that'll be used to identify a chunk. 15:36
no name wouldn't be valid 15:37
an empty name would be fine (though odd)
I don't care too much about quotes, as long as it's unsurprising 15:38
colliding names would be bad
dukeleto cotto: sure 15:39
cotto: let's make the spec simple and just require one kind of quote 15:40
cotto: i don't want to support unicode quotes or anything like that
cotto: you ok with requiring double quotes? 15:41
cotto dukeleto, wfm 15:43
dalek nxed: r980 | NotFound++ | trunk/winxedst1.winxed:
use Expected and Unexpected in apropiate places instead of the generic
 15:44
rrot/m0-prototype: 9e83f7d | dukeleto++ | t/m0/ (3 files):
Add tests for parsing invalid chunk names
 15:46
cotto is out until this evening. 15:48
15:53 woosley left 15:57 benabik joined
dalek rrot/m0-spec: 4d8e965 | dukeleto++ | docs/pdds/draft/pdd32_m0.pod:
Add more detail about chunk identifiers
 15:59
16:02 lateau1 left
dalek nxed: r981 | NotFound++ | trunk/winxedst1.winxed:
simplify parseList usage
 16:23
rrot: 574bb32 | luben++ | src/gc/gc_gms.c:
new defaults for GC GMS - they give arround 10% speedup
 16:38
rrot: 0746879 | luben++ | / (7 files):
add --gc-nursery-size runtime parameter for parrot
16:42 davidfetter joined 16:58 theory joined
dalek rrot: cbf6649 | NotFound++ | src/gc/gc_gms.c:
cast to size_t instead of int for g++ compatibility and deleted unused local vars
 17:12
rrot: fc6299d | NotFound++ | frontend/parrot/main.c:
fix fprintf format
 17:13
17:14 contingencyplan joined 17:49 rurban_ joined
dalek rrot/m0-prototype: 4179b8e | dukeleto++ | t/m0/ (2 files):
Add an M0 test file that has an empty chunk name and relevant tests
 17:51
17:51 rurban left 17:52 rurban_ is now known as rurban 18:04 kid51 joined
nopaste "kid51" at 192.168.1.3 pasted "t/pmc/callcontext.t: new failure on master on linux/i386" (83 lines) at nopaste.snit.ch/44037 18:05
kid51 If I read the commit log for the last day, those failures are predictable. 18:08
darwin/PPC: builds successfully on master at HEAD with --cc=gcc ... so it's the attempt to build with --cc=g++ that is running into GC problems. 18:10
18:24 birdwindupbird left
dalek rrot/m0-prototype: 5895715 | dukeleto++ | / (2 files):
Add a test for the edge case of a chunk name that is the empty string; Clean up generated .m0b files
 18:32
dukeleto kid51: that is expected, bacek++ broke it on purpose. Look at the /topic
nopaste "kid51" at 192.168.1.3 pasted "darwin/ppc: master: two test failures" (19 lines) at nopaste.snit.ch/44038 18:39
dalek rrot/m0-prototype: 7c764be | dukeleto++ | / (2 files):
Fix a bug in our M0 parsing and modify one of the M0 test files to verify it is fixed
kid51 Those darwin/ppc failures were with cc=gcc. 18:40
But that failure message ... PARROT VM: Could not initialize new interpreterunimplemented value type ... I got during 'make' with cc=g++ 18:41
18:41 kid51 left 18:42 TiMBuS joined 18:47 theory left
dalek rrot: b45e5f3 | plobsing++ | src/pmc/callcontext.pmc:
clone pcc cells to avoid double-free issues
 18:52
plobsing msg bacek does b45e5f3 satisfy the issue you found? 18:54
aloha OK. I'll deliver the message.
18:58 whiteknight joined
dalek rrot: 0794278 | plobsing++ | / (31 files):
Merge branch 'tt1931-nci-parameters-deprecation'
 19:07
19:14 theory joined
dalek rrot: 9767706 | NotFound++ | tools/dev/pbc_to_exe.pir:
add a few newlines to avoid ugly looking messages
 19:19
19:31 hudnix joined 19:32 dodathome left
dalek rrot: 11483b2 | NotFound++ | src/hash.c:
improve some exceptions messages and types in hash.c
 19:35
19:44 Coke left 19:50 Coke joined
dalek rrot: b71eb28 | plobsing++ | src/pmc/callcontext.pmc:
[codingstd] c_macro_args
 19:51
rrot: e34c654 | plobsing++ | src/pmc/callcontext.pmc:
[codingstd] c_parens
rrot: 2dd3f75 | plobsing++ | / (3 files):
[codingstd] trailing_space
rrot: 7fae0c7 | plobsing++ | / (2 files):
Merge branch 'master' of github.com:parrot/parrot
19:57 Coke left 19:58 Coke joined
NotFound $ ./parrot examples/opengl/triangle.pir 20:01
Segmentation fault
whiteknight ...it's a feature 20:09
NotFound Did we have some nci documentation that's not a bag of lies? 20:14
20:15 ambs_ joined 20:19 bluescreen joined, ambs left, ambs_ is now known as ambs
NotFound Nice, I can't run any example in examples/nci 20:26
20:28 Coke left, Coke joined
bubaflub NotFound: i don't know if it's related, but plobsing merged tt1931-nci-parameters-deprecation about an hour and a half ago at github.com/parrot/parrot/commit/0794278e5c 20:30
NotFound bubaflub: I'm pretty sure is related. That doesn't help. 20:33
20:36 Coke left, Coke joined
whiteknight apparently Richard Stallman is publishing two new books 20:37
tadzik oh, source?
whiteknight surprisingly, the titles are not "HUBAGRABAWHARBLEGARBLE" and "Goverment Did 9/11"
www.fsf.org/news/rms-newbooks
tadzik shame, the first one seems sane 20:39
I think I read Free as in Freedom
NotFound What the fuck 't - manage null-padding in sugar layer' is supposed to mean? 20:42
ttbot Parrot 0794278e MSWin32-x86-multi-thread make error tt.taptinder.org/cmdinfo/1016 20:44
lucian_ whiteknight: i love how they both have his hairy face on them
he'd be great marketing if he were more pragmatic
20:44 lucian_ is now known as lucian
whiteknight lucian: yeah, on one hand I really admire some of the things he did and starting the open source movement and all 20:45
lucian whiteknight: correction, OSI is someone's else brainchild
but yeah, i do admire most of what he did
whiteknight on the other hand, he's gotten so much more absurd, fundamentalist, and crap
lucian on a totally unrelated note, i rather like Unity 20:46
whiteknight like, I saw a speech where he suggested in an off-hand way that the government was a conspirator in 9/11
that's not really an opinion that exudes "sanity"
lucian whiteknight: he really did that? that's a bit odd
whiteknight a bit
I would have to find a link
His fights with Linux Torvalds, because he isn't "concerned about freedom" are absurd 20:47
lucian i mean sure, there are odd things about 9/11. i can understand a skeptical mind not being convinced entirely that there wasn't more at play than what was seen
whiteknight his criticisms of Miguel de Icaza for the same reasons are also absurd
lucian but rms of all people? i thought he only cared about software
lucian nods
whiteknight apparently he doesn't write any software anymore
he only criticizes the people who do write the software 20:48
lucian i'd like to see a GPL 3 without the anti-tivoisation, and something like AGPL that added anti-tivoisation
whiteknight isn't that the GPL2?
:)
lucian no, GPL 3 has a few added niceties
besides the tivo bit, it's overall better than gpl2
and i find it very sad that rms does have a point about mono 20:49
it was, and still is, a good idea to avoid relying on it
what's happening to novell now only goes to reinforce that
20:49 theory left
NotFound I switched from Suse when Novell bought it. 20:50
lucian shrugs 20:51
i still used mono for various things at various times, i wouldn't say it's useless
but potentially dangerous? sure
i just pains me rms is so extremist
NotFound I uninstalled libmono from my systems and don't have any problem. 20:52
lucian burrows back to his hole, where the dreaded dissertation still awaits, in the dark, breathing heavily
NotFound C# is not bad language, though.
lucian NotFound: true, it's much better than Java 20:53
tadzik there is this scientic term, the frame of somebody I think
20:53 bubaflub left
lucian NotFound: which is why sort of like code.google.com/p/stab-language/ 20:53
tadzik so because RMS is weird, Linus does not seem that weird anymore, that kinda thing
not that I think Linus is weird, that's just an example 20:54
lucian tadzik: i'd say they're both weird, tbh
linus just a bit, though
plobsing if you want to use C# without libmono, you could use dotgnu. 20:55
NotFound A lot of people spend his time telling the world the weirdness of other people.
plobsing anyone else having track login problems? 20:56
lucian plobsing: that's no help, sucks even wose
plobsing lucian: how does it suck? I've not tried it, just heard of its existance.
ttbot Parrot 7fae0c78 MSWin32-x86-multi-thread make error tt.taptinder.org/cmdinfo/1028
lucian plobsing: it's extremely incomplete and has been abandoned
NotFound I don't think is a bad language, but I don't have special need to use it.
lucian plobsing: there isn't anything inherently wrong with mono, just with .net implementations in general 20:57
NotFound The problem is that no one wants to use CLI, all users want full .net
So the 'C' in CLI means almost nothing. 20:58
benabik I tend to like Scala these days, which runs on CLI and JVM.
plobsing that's not a technical problem, but a social one. don't hate the implementation, hate the community. 20:59
NotFound I don't hate, just don't use.
Hate gives you to the Dark Side.
lucian NotFound has a valid point. even mono is lagging between .net 21:01
if you target mono, it's easy to have it portable to .net
but it doesn't quite work the other way around, even ignoring all the gratuitous native calls
plobsing from the people behind embrace and extend, comes CLI! suprised? 21:02
dalek rrot: 6dd6103 | plobsing++ | src/libnci_test.def:
remove unimpelemented export
 21:04
lucian their "common" means it's similar enough to the jvm that you can use other languages than C# on it
plobsing if you could truly do that, VB.Net wouldn't have been nearly so incompatible 21:05
whiteknight tadzik: Linus can be a huge jerk sometimes, in addition to his fanaticism 21:07
linus' fanaticism is often profanity-laced and humerous, unlike rms' which is crazy and depressing
plobsing: yes, I cannot log into trac 21:08
plobsing fun times. I had hoped to close out this deprecation today.
whiteknight I keep thinking we should move away from trac. So far, none of the alternatives are less bad than trac 21:09
trac is plenty bad, but less bad than the alternatives
I think the github issue tracker is not too far away from something that would be suitable, but it's just not there yet 21:10
plobsing sd would be awesome if it weren't beta-quality
whiteknight yeah
lucian whiteknight: trac's issue tracker is pretty good, just trac in general is a tad bitrotten
it's easy to make a good one, i don't get why github are dragging their asses
plobsing have the appropriate people been pinged about our trac login problems? who are the appropriate people? 21:11
benabik Github's issue tracker is better than a non-functional trac. 21:12
lucian there's launchpad, they have a decent issue tracker 21:14
plobsing the issue is that decent isn't good enough. to justify the effort of migrating all of our trac info to something else, we need guarrantees that it is better in pretty much every way, satisfies our outstanding issues with trac, and has a handful of volunteers eager to perform the lion's share of the migration work. 21:16
ttbot Parrot 6dd6103f MSWin32-x86-multi-thread make error tt.taptinder.org/cmdinfo/1048 21:17
lucian plobsing: lp is likely better in every way than trac 21:18
21:20 SHODAN left
benabik lucian: Other than the fact that lp is integrated with bazaar rather than git. 21:22
lucian benabik: is trac integrated with git?
benabik lucian: IIRC, trac.parrot.org has a git plugin of some sort. 21:23
plobsing lucian: trac has a git plugin
lucian i see
NotFound I've read several javascript proposed enhancements and looks like foolish enhanced featurism to me. 21:24
benabik What does github issue tracing not have that we use in trac?
lucian how deep does the trac-git integration go?
benabik: try using it for a bit, it's actually funny 21:25
it's essentially a forum, where you can add tags to posts
much less than even that, in fact
21:26 ambs left
benabik Lack of metadata is the issue? 21:27
whiteknight plobsing: we were in contact with OSU yesterday. I don't know the current status of the issue 21:28
lucian benabik: mostly, yes. it barely has any at all 21:29
21:30 bubaflub joined
lucian benabik: have a look github.com/jashkenas/coffee-script/issues 21:30
whiteknight from what I just read, the issue with OSU is unresolved. I just sent them another email 21:31
whiteknight is leaving for now, will be back later for status
dalek rrot: e58450b | plobsing++ | / (2 files):
accomodate libnci_test not being linked against libparrot
 21:36
benabik lucian: I see it as "different", not "lacking". I don't see a big difference in labels vs static metadata categories. (esp when I've seen several issue DBs with very very lax usage of the metadata) 21:41
lucian benabik: i find features such as file upload, ticket states, assignment and so on very useful, so i disagree 21:42
they've improved it lately though, it used to be even worse
plobsing why would you need file upload? you have pull requests to replace patches and gists to replace other text uploads. </github-perspective> 21:44
lucian plobsing: perhaps. i tend to see patches and pull requests as complementary
benabik lucian: They do have assignment and milestones. Non-open/closed status is a label, and file uploads are probably expected to be gist or pull requests... Eh. Different, but not really worse.
lucian: IMO, of course. 21:45
lucian benabik: sure. i don't think it's much worse than trac is right now
but i do think there's to be had better (bitbucket, launchpad)
benabik lucian: And the interface is far more welcoming to users. The fact that trac presents all ticket options for editing when you want to post a comment is irritating at best. 21:46
lucian benabik: true
i do think that ticket states should be separate from general labels
and a few other nits
benabik (Confusing, off-putting, and dangerous at worst.)
lucian and i really think it needs an assignment feature
benabik lucian: It has issue assignment. 21:47
lucian benabik: to users? i must've missed that
benabik lucian: Yup. Showed up in "Issues 2.0"
plobsing if you want to talk offputting UI, I have UI issues with launchpad. I can never figure out where the info I want is when I'm checking out a hosted project.
lucian i see
plobsing: yes, the ui is odd
overall, i think github > launchpad 21:48
even with git, and even with inferior issue tracking
benabik shrugs.
lucian benabik: it's a good thing to be able to do that consistently :D 21:49
lucian shrugs a lot too
22:34 fperrad left
dalek rrot: 4e63bc3 | mikehh++ | src/nci_test.c:
correct =item details
 22:46
23:03 bacek left 23:27 bacek joined 23:38 theory joined
whiteknight if github's issue tracker was git-based, so we could download and play with local copies, that would be a feature that would push it over the edge for me 23:52
github does have the ability to add labels to issues, but that's limited. Some tickets really need the ability to add lots of metadata 23:53
of course, many tickets don't need too much metadata

Powered by Cro and the Raku® Programming Language.

Please report any issues / comments / feature requests as an issue on App::Raku::Log.

Thank you!