Switch to DuckDuckGo Search
   June 2, 2019
< | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 30

Toggle Join/Part | bottom
[00:07:12] *** baojg <baojg!~baojg@162.243.44.213> has joined #illumos
[00:12:24] *** baojg <baojg!~baojg@162.243.44.213> has quit IRC (Ping timeout: 272 seconds)
[00:22:34] *** merzo <merzo!~merzo@206-8-132-95.pool.ukrtel.net> has quit IRC (Read error: Connection reset by peer)
[00:23:45] *** merzo <merzo!~merzo@206-8-132-95.pool.ukrtel.net> has joined #illumos
[00:30:04] *** CVLTCMK0 <CVLTCMK0!~Ident@ool-2f1257d8.dyn.optonline.net> has quit IRC (Quit: ZNC - http://znc.in)
[00:50:16] *** CVLTCMK0 <CVLTCMK0!~Ident@ool-2f1257d8.dyn.optonline.net> has joined #illumos
[01:15:41] *** andy_js <andy_js!~andy@94.14.32.34> has quit IRC (Quit: andy_js)
[01:32:16] *** mnrmnaugh <mnrmnaugh!~mnrmnaugh@unaffiliated/mnrmnaugh> has quit IRC (Remote host closed the connection)
[01:33:15] *** mnrmnaugh <mnrmnaugh!~mnrmnaugh@unaffiliated/mnrmnaugh> has joined #illumos
[02:08:00] *** baojg <baojg!~baojg@162.243.44.213> has joined #illumos
[02:12:16] *** baojg <baojg!~baojg@162.243.44.213> has quit IRC (Ping timeout: 248 seconds)
[02:45:31] *** btibble <btibble!~brantibbl@c-69-94-200-89.hs.gigamonster.net> has quit IRC (Ping timeout: 244 seconds)
[03:20:00] *** ed209 <ed209!~ed209@165.225.128.67> has quit IRC (Remote host closed the connection)
[03:20:09] *** ed209 <ed209!~ed209@165.225.128.67> has joined #illumos
[04:06:45] *** CVLTCMK0 <CVLTCMK0!~Ident@ool-2f1257d8.dyn.optonline.net> has quit IRC (Quit: ZNC - http://znc.in)
[04:08:51] *** baojg <baojg!~baojg@162.243.44.213> has joined #illumos
[04:13:23] *** baojg <baojg!~baojg@162.243.44.213> has quit IRC (Ping timeout: 258 seconds)
[04:43:12] *** mnrmnaugh <mnrmnaugh!~mnrmnaugh@unaffiliated/mnrmnaugh> has quit IRC (Remote host closed the connection)
[04:44:05] *** mnrmnaugh <mnrmnaugh!~mnrmnaugh@unaffiliated/mnrmnaugh> has joined #illumos
[04:59:29] *** baojg <baojg!~baojg@162.243.44.213> has joined #illumos
[05:55:44] *** DrLou <DrLou!~DrLou@c-73-33-92-169.hsd1.nj.comcast.net> has quit IRC (Ping timeout: 248 seconds)
[06:01:52] *** DrLou <DrLou!~DrLou@c-73-33-92-169.hsd1.nj.comcast.net> has joined #illumos
[06:06:17] *** DrLou <DrLou!~DrLou@c-73-33-92-169.hsd1.nj.comcast.net> has quit IRC (Ping timeout: 252 seconds)
[06:07:09] *** baojg <baojg!~baojg@162.243.44.213> has quit IRC (Remote host closed the connection)
[06:28:36] *** jelmd <jelmd!~purple@p5DC04845.dip0.t-ipconnect.de> has quit IRC (Ping timeout: 272 seconds)
[06:38:59] *** baojg <baojg!~baojg@162.243.44.213> has joined #illumos
[06:58:31] *** freakazoid0223 <freakazoid0223!~matt@pool-108-52-4-3.phlapa.fios.verizon.net> has quit IRC (Remote host closed the connection)
[07:10:14] *** BOKALDO <BOKALDO!~BOKALDO@46.109.203.201> has joined #illumos
[08:25:34] *** ibenn <ibenn!~benn@HSI-KBW-095-208-237-104.hsi5.kabel-badenwuerttemberg.de> has joined #illumos
[08:57:55] *** jthan <jthan!~jonathan@maxilaria.jrdemasi.com> has quit IRC (Ping timeout: 246 seconds)
[09:04:57] *** jthan <jthan!~jonathan@maxilaria.jrdemasi.com> has joined #illumos
[09:07:11] *** phyre__ <phyre__!~phyre___@31.211.191.164> has joined #illumos
[09:38:33] *** andy_js <andy_js!~andy@94.14.32.34> has joined #illumos
[10:04:54] *** ibenn <ibenn!~benn@HSI-KBW-095-208-237-104.hsi5.kabel-badenwuerttemberg.de> has quit IRC (Quit: Leaving)
[10:11:32] *** phyre__ <phyre__!~phyre___@31.211.191.164> has quit IRC (Ping timeout: 272 seconds)
[10:32:46] *** Diagon <Diagon!DiagonalAr@gateway/vpn/protonvpn/diagonalarg> has quit IRC (Ping timeout: 246 seconds)
[10:38:51] *** baojg <baojg!~baojg@162.243.44.213> has quit IRC (Remote host closed the connection)
[10:39:25] *** baojg <baojg!~baojg@162.243.44.213> has joined #illumos
[10:46:12] *** Diagon <Diagon!DiagonalAr@gateway/vpn/protonvpn/diagonalarg> has joined #illumos
[11:10:10] *** amrfrsh <amrfrsh!~Thunderbi@95.174.67.172> has quit IRC (Quit: amrfrsh)
[11:57:07] *** jeffpc <jeffpc!~jeffpc@josefsipek.net> has quit IRC (Ping timeout: 246 seconds)
[12:14:39] *** amrfrsh <amrfrsh!~Thunderbi@131.234.44.147> has joined #illumos
[12:17:18] *** keremet <keremet!~keremet2@92.255.237.50> has joined #illumos
[12:20:00] *** ed209 <ed209!~ed209@165.225.128.67> has quit IRC (Remote host closed the connection)
[12:20:07] *** ed209 <ed209!~ed209@165.225.128.67> has joined #illumos
[12:20:31] *** DrLou <DrLou!~DrLou@c-73-33-92-169.hsd1.nj.comcast.net> has joined #illumos
[12:42:38] <sjorge> andy_js and co also sort of aware of KEA... https://gitlab.isc.org/isc-projects/kea/issues/631#note_60306
[12:42:45] <sjorge> Dod you see the note about our if/net.h ?
[13:37:01] *** jlevon <jlevon!~jlevon@movementarian.org> has quit IRC (Quit: Coyote finally caught me)
[13:38:36] *** jlevon <jlevon!~jlevon@movementarian.org> has joined #illumos
[13:53:29] *** jelmd <jelmd!~purple@p5DC04EB8.dip0.t-ipconnect.de> has joined #illumos
[13:53:46] *** ptribble <ptribble!~ptribble@cpc92716-cmbg20-2-0-cust138.5-4.cable.virginm.net> has joined #illumos
[14:15:14] *** hemi770 <hemi770!~hemi666@unaffiliated/hemi770> has quit IRC (Ping timeout: 258 seconds)
[14:16:48] *** hemi770 <hemi770!~hemi666@unaffiliated/hemi770> has joined #illumos
[14:44:12] *** clapont <clapont!~clapont@unaffiliated/clapont> has quit IRC (Ping timeout: 244 seconds)
[14:46:26] *** clapont <clapont!~clapont@unaffiliated/clapont> has joined #illumos
[14:58:38] *** mno-hime <mno-hime!~mno-hime@94.142.238.232> has quit IRC (Quit: Leaving)
[15:01:26] *** mno-hime <mno-hime!~mno-hime@94.142.238.232> has joined #illumos
[15:11:46] *** keremet <keremet!~keremet2@92.255.237.50> has left #illumos
[15:20:55] *** idodeclare <idodeclare!~textual@104-189-74-222.lightspeed.snantx.sbcglobal.net> has joined #illumos
[16:55:02] *** freakazoid0223 <freakazoid0223!~matt@pool-108-52-4-3.phlapa.fios.verizon.net> has joined #illumos
[17:29:37] *** alanc <alanc!~alanc@129.157.69.41> has quit IRC (Remote host closed the connection)
[17:30:06] *** alanc <alanc!~alanc@129.157.69.41> has joined #illumos
[17:30:34] *** amrfrsh <amrfrsh!~Thunderbi@131.234.44.147> has quit IRC (Quit: amrfrsh)
[17:32:10] <gitomat> [illumos-gate] 10210 libmd should leverage SHA extensions -- Robert Mustacchi <rm at joyent dot com>
[17:32:45] <andyf> richlowe - ping
[17:35:28] *** merzo <merzo!~merzo@206-8-132-95.pool.ukrtel.net> has quit IRC (Ping timeout: 248 seconds)
[17:38:16] <andyf> We have a strange regression in libm after 10881
[17:39:05] *** merzo <merzo!~merzo@206-8-132-95.pool.ukrtel.net> has joined #illumos
[17:43:23] <andyf> 32-bit perl with `-MPOSIX -e 'print tanh(-1)'` is giving the same result as tanh(1) - wrong sign
[17:44:13] <andyf> If I removed the new define for signbit() and rebuild libm, it works properly
[17:44:18] <andyf> *remove
[18:21:15] *** baojg <baojg!~baojg@162.243.44.213> has quit IRC (Remote host closed the connection)
[18:29:39] *** ptribble <ptribble!~ptribble@cpc92716-cmbg20-2-0-cust138.5-4.cable.virginm.net> has quit IRC (Quit: Leaving)
[18:42:21] *** clapont <clapont!~clapont@unaffiliated/clapont> has quit IRC (Ping timeout: 252 seconds)
[18:44:00] *** clapont <clapont!~clapont@unaffiliated/clapont> has joined #illumos
[18:49:04] *** clapont <clapont!~clapont@unaffiliated/clapont> has quit IRC (Ping timeout: 246 seconds)
[18:51:12] *** clapont <clapont!~clapont@unaffiliated/clapont> has joined #illumos
[18:58:51] *** btibble <btibble!~brantibbl@c-69-94-200-89.hs.gigamonster.net> has joined #illumos
[19:04:41] *** phyre__ <phyre__!~phyre___@31.211.191.164> has joined #illumos
[19:11:08] *** merzo <merzo!~merzo@206-8-132-95.pool.ukrtel.net> has quit IRC (Ping timeout: 272 seconds)
[19:13:48] *** phyre__ <phyre__!~phyre___@31.211.191.164> has quit IRC (Ping timeout: 245 seconds)
[19:27:09] *** merzo <merzo!~merzo@253-53-133-95.pool.ukrtel.net> has joined #illumos
[19:29:12] *** MilkmanDan <MilkmanDan!~dan@wilug/expat/MilkmanDan> has quit IRC (Ping timeout: 245 seconds)
[19:31:15] *** MilkmanDan <MilkmanDan!~dan@wilug/expat/MilkmanDan> has joined #illumos
[19:43:36] <tsoome> but 64-bit one?
[19:43:47] *** MilkmanDan <MilkmanDan!~dan@wilug/expat/MilkmanDan> has quit IRC (Ping timeout: 245 seconds)
[19:44:49] *** MilkmanDan <MilkmanDan!~dan@wilug/expat/MilkmanDan> has joined #illumos
[20:22:07] *** baojg <baojg!~baojg@162.243.44.213> has joined #illumos
[20:23:11] *** amrfrsh <amrfrsh!~Thunderbi@95.174.67.172> has joined #illumos
[20:24:53] *** amrfrsh <amrfrsh!~Thunderbi@95.174.67.172> has quit IRC (Client Quit)
[20:26:43] *** baojg <baojg!~baojg@162.243.44.213> has quit IRC (Ping timeout: 245 seconds)
[20:31:26] *** danmcd <danmcd!~danmcd@static-71-174-113-16.bstnma.fios.verizon.net> has joined #illumos
[20:44:39] *** amrfrsh <amrfrsh!~Thunderbi@95.174.67.172> has joined #illumos
[20:45:33] *** amrfrsh <amrfrsh!~Thunderbi@95.174.67.172> has quit IRC (Client Quit)
[21:13:08] *** amrfrsh <amrfrsh!~Thunderbi@95.174.67.172> has joined #illumos
[21:16:38] *** XV8 <XV8!~XV8@d-146-168-23-31.md.cpe.atlanticbb.net> has joined #illumos
[21:26:27] <andyf> tsoome - 64-bit is fine
[21:26:44] <andyf> and if I write my own small C program using libm, that seems fine in 32-bit too
[21:26:51] <andyf> Does OI have 32-bit perl still?
[21:28:55] <toasterson> andyf yes
[21:29:08] <toasterson> we ship both
[21:29:21] <toasterson> one is 5.22 the other 5.24
[21:29:27] <andyf> Can you see if you have the same issue?
[21:29:51] <andyf> ISALIST=i386 perl -MPOSIX -e 'print tanh(-1) . "\n"'
[21:29:55] <andyf> or similar
[21:30:13] <toasterson> sure
[21:30:14] <tsoome> 0.761594155955765
[21:31:26] * toasterson sent a long message: < http://192.168.10.17:8008/_matrix/media/v1/download/matrix.wegmueller.it/obgZvlCMUoyaDzrbZiaXOOuA >
[21:31:59] <andyf> yeah.. it should be negative
[21:32:01] <andyf> try it with 64-bit
[21:32:18] <toasterson> those are both variants of perl
[21:32:32] <toasterson> do i need to change ISALIST?
[21:32:38] <andyf> toasterson - I can't get to that link..
[21:32:54] <andyf> It's 192.168.10....
[21:33:11] <toasterson> ah thats the problem with my homeserver :)
[21:33:13] <LeftWing> toasterson: I guess you're using Matrix as an IRC client, and your instance is hosted on a private network.
[21:33:22] <LeftWing> You'll need to gist long messages.
[21:33:27] <toasterson> could you post the link so i can fix it?
[21:33:35] <andyf> http://192.168.10.17:8008/_matrix/media/v1/download/matrix.wegmueller.it/obgZvlCMUoyaDzrbZiaXOOuA
[21:33:45] <toasterson> yes i didn't know what long means for matrix
[21:34:35] *** DrLou <DrLou!~DrLou@c-73-33-92-169.hsd1.nj.comcast.net> has quit IRC (Quit: DrLou)
[21:35:21] <toasterson> https://gist.github.com/Toasterson/b4eca9001bead69e248afe1ff42ed82d
[21:38:24] <andyf> tsoome - was yours the same command? toasterson got the right answers.
[21:42:24] <tsoome> /usr/perl5/5.24/bin/perl5.24.3 -MPOSIX -e 'print tanh(-1) . "\n"'
[21:43:18] <tsoome> v5.22.4 is broken
[21:44:11] <toasterson> oi only has 5.24.3
[21:45:17] <tsoome> I have 5.22 for some reason
[21:45:50] <toasterson> we have both 🙂 in oi
[21:46:00] <toasterson> ones 32bit ones 64bit
[21:47:44] <andyf> It's libm.so.2 that's broken
[21:47:58] <andyf> I just don't know exactly how yet
[21:48:39] *** phyre__ <phyre__!~phyre___@31.211.191.164> has joined #illumos
[21:52:50] <andyf> Well, something to do with the gcc4 builtin for signbit() in 32-bit, but only affecting the tanh() function as far as I can tell
[22:00:21] *** BOKALDO <BOKALDO!~BOKALDO@46.109.203.201> has quit IRC (Quit: Leaving)
[22:12:58] *** phyre__ <phyre__!~phyre___@31.211.191.164> has quit IRC (Ping timeout: 245 seconds)
[22:17:09] *** krzysztofkanas <krzysztofkanas!~chris@176-111-123-228.net.cybernetwmw.com> has joined #illumos
[22:20:01] *** ed209 <ed209!~ed209@165.225.128.67> has quit IRC (Remote host closed the connection)
[22:20:08] *** ed209 <ed209!~ed209@165.225.128.67> has joined #illumos
[22:22:59] *** baojg <baojg!~baojg@162.243.44.213> has joined #illumos
[22:26:14] *** DrLou <DrLou!~DrLou@c-73-33-92-169.hsd1.nj.comcast.net> has joined #illumos
[22:27:44] *** baojg <baojg!~baojg@162.243.44.213> has quit IRC (Ping timeout: 248 seconds)
[22:32:54] <neirac> toasterson have you had time to check this pr? https://git.wegmueller.it/Illumos/go-zone/pulls/2
[22:33:15] <toasterson> yes can't you see the comments?
[22:34:42] <neirac> toasterson no, I don't see any comment
[22:35:18] <toasterson> not even under the changed files tab?
[22:36:14] *** XV9 <XV9!~XV8@d-146-168-23-31.md.cpe.atlanticbb.net> has joined #illumos
[22:37:00] <neirac> toasterson no, I don't see anything
[22:37:21] <toasterson> ??? let me check without account
[22:39:03] <toasterson> ah i needed to press one more button than github
[22:39:05] *** XV8 <XV8!~XV8@d-146-168-23-31.md.cpe.atlanticbb.net> has quit IRC (Ping timeout: 268 seconds)
[22:39:15] <toasterson> should see it now
[22:40:47] *** merzo <merzo!~merzo@253-53-133-95.pool.ukrtel.net> has quit IRC (Read error: Connection reset by peer)
[22:42:43] *** merzo <merzo!~merzo@253-53-133-95.pool.ukrtel.net> has joined #illumos
[22:56:36] *** clapont <clapont!~clapont@unaffiliated/clapont> has quit IRC (Ping timeout: 272 seconds)
[23:06:40] *** joltman <joltman!znc@gateway/vpn/privateinternetaccess/joltman> has quit IRC (Ping timeout: 246 seconds)
[23:06:54] *** sarvet <sarvet!~sarvet@dslb-092-073-092-141.092.073.pools.vodafone-ip.de> has joined #illumos
[23:10:47] <andyf> toasterson / tsoome - opened https://www.illumos.org/issues/11175 with C reproducer rather than relying on perl
[23:13:32] *** joltman <joltman!~znc@unaffiliated/joltman> has joined #illumos
[23:13:52] <toasterson> andyf what should the result be actually
[23:14:09] <andyf> tanh(-1.000000) = -0.761594
[23:14:10] <andyf> tanh(1.000000) = 0.761594
[23:14:36] <toasterson> is it wrong on todays gate or in a pr?
[23:14:42] <andyf> It's wrong in today's gate
[23:15:04] <toasterson> ah. we haven't gotten it yet untill in 45 minutes + buildtime
[23:15:22] <andyf> It became broken with 10881 which was integrated 4 weeks ago
[23:15:45] <toasterson> ? then why is it working with up to date OI?
[23:16:19] <andyf> It wasn't for tsoome.. maybe OI doesn't use isaexec for perl in the way I expected
[23:16:24] <andyf> try the C program in the issue
[23:16:31] <toasterson> i did
[23:16:44] <toasterson> thats why i was asking what the result should be
[23:17:17] <andyf> Does the OI build server get updated headers or could it still have an old math_c99.h?
[23:17:58] <toasterson> that could be the case yes..... shouldn't it use the headers from gate?
[23:18:24] <andyf> yes, it should, forget my last comment
[23:18:48] <toasterson> well the problem is the word should.....
[23:19:00] <toasterson> it could be broken even more....
[23:19:06] <andyf> well, should and as long as the proto area is cleaned out
[23:20:11] <toasterson> that would be our problem on the build server.... but i hope it is. leoric or mno-hime would need to share some insight into this.
[23:21:52] <richlowe> andyf: is that where I changed stuff to use the builtins?
[23:22:03] <andyf> richlowe - yes
[23:22:15] <andyf> in particular, __builtin_signbit() seems to cause the problem
[23:22:37] <toasterson> added a comment with my result
[23:22:41] <richlowe> signbit is also the thing that caused me to change it :\
[23:24:10] <richlowe> andyf: does float v. double matter?
[23:24:47] *** clapont <clapont!~clapont@unaffiliated/clapont> has joined #illumos
[23:25:22] <andyf> no, but if I just do "tanh(-1)" in C, it works
[23:25:32] <andyf> just not tanh((float)-1) or tanh((double)-1)
[23:26:37] <andyf> actually, explicit casts do work
[23:27:59] <andyf> Here - https://paste.ec/paste/YsRMzaqp#6gTJXo3KlUC2KZn4JEq9NEbw4IBnjupDR0LfhcVveyZ
[23:30:33] <richlowe> the difference is because the constants will use the builtin tanh without -fno-builtin
[23:31:20] <richlowe> otherwise, when 32bit (only) tanh flips the sign.
[23:31:31] <richlowe> is that what you see?
[23:32:07] <richlowe> gcc -fno-builtin math.c -o math -lm # flips the sign
[23:32:14] <richlowe> gcc -fno-builtin math.c -m64 -o math -lm # doesn't.
[23:32:30] <richlowe> actually, not flips, it's always positive
[23:33:08] <andyf> with -fno-builtin, I always get a positive result
[23:33:14] <richlowe> 64bit _and_ 32bit?
[23:33:16] <andyf> (i.e. wrong)
[23:33:26] <andyf> no, 64-bit is correct
[23:33:39] <richlowe> now to work out which tanh i386 uses
[23:33:46] <andyf> so yes, that's what I'm seeing
[23:34:49] <richlowe> andyf: ask advocates@ to back me out, I guess?
[23:36:30] <richlowe> I have a guess at what's going on, but it's not a good guess.
[23:36:35] <andyf> I suppose so..
[23:36:39] <richlowe> actually, it's an amazingly good guess
[23:36:53] <richlowe> signbit() returns _non-zero_ if negative
[23:36:55] <richlowe> tanh() compares it to 1
[23:37:02] <richlowe> while 1 is not non-0, it is far from the set of non-zero integers.
[23:37:13] <andyf> It isn't affecting any supported OmniOS release so no rush from me
[23:37:26] <richlowe> andyf: so we get into back compatibility with broken code
[23:37:30] <andyf> so forward to a fix, or revert, no preference
[23:37:48] <andyf> richlowe - I wonder why 64-bit is ok
[23:38:06] <andyf> I looked at our own gate implementation of signbit(). In 32-bit it just returns the top word >> 31
[23:38:09] <richlowe> signbit() returns 1 there
[23:38:10] <andyf> so always 1 or 0
[23:38:30] <richlowe> so we could fix the macro to be (__builtin_signbit(x) > 0)
[23:39:15] <richlowe> looking for other broken code
[23:39:45] <richlowe> tanh() and jn() are broken
[23:39:47] <richlowe> nothing else seems to be
[23:40:01] <andyf> jn() ?
[23:40:07] <richlowe> I'm not a math nerd, dunno
[23:40:08] <richlowe> it's a math thing!
[23:40:16] <richlowe> ask alarcher! :)
[23:40:23] <richlowe> put that professor thing to use
[23:40:27] <andyf> if (sgn == 1)
[23:40:27] <andyf> return (-b);
[23:40:34] <richlowe> precisely
[23:40:37] *** idodeclare <idodeclare!~textual@104-189-74-222.lightspeed.snantx.sbcglobal.net> has quit IRC (Quit: My MacBook has gone to sleep. ZZZzzz…)
[23:40:37] <andyf> where sng = signbit(), so looks broken
[23:40:44] <andyf> that's C/jn.c
[23:40:46] <igork> andyf: https://paste.dilos.org/?2e12e56c0dcef88d#TT4Cq6mS+C16zYzGGvdr7HUJozGOxTbZnEECVmxadX0=
[23:40:47] <richlowe> andyf: could you email advocates@ so we can start the compat conversation?
[23:41:09] <andyf> Broken in Dilos too?!
[23:41:12] <andyf> richlowe - will do
[23:41:19] <richlowe> andyf: I imagine the fix I proposed above where signbit(x) becomes (__builtin_signbit(x) > 0) is a good noe
[23:41:21] <igork> andyf: just checked on dilos by gcc6
[23:41:34] <richlowe> andyf: if you wanted to check that quickly
[23:42:15] <andyf> testing now...
[23:42:17] <richlowe> andyf: does however you found this cover the other things I touched? and what was it?
[23:42:28] <richlowe> I'd like to note it down for any future such changes, so it doesn't happen again.
[23:42:40] <richlowe> and I'm sorry, I didn't realize our libm was busted this way :(
[23:42:46] <andyf> It was @hadfl - he was running the perl testsuite and it found this when testing the POSIX module
[23:42:59] <richlowe> looks like in illumos only libm is broken
[23:43:00] <richlowe> the irony
[23:43:08] <andyf> yes, (__builtin_signbit(x) > 0) fixes it
[23:43:23] <igork> andyf: with gcc version?
[23:43:26] <richlowe> I would _expect_ 3rd party code to be immune, since they likely used the builtin on some platforms regardless.
[23:43:38] <andyf> igork - with the illumos gcc4.4.4
[23:43:45] <richlowe> but with such a simple fix it's probably best we do
[23:44:00] <richlowe> andyf: would the perl tests have found other such problems, were there any?
[23:44:44] <andyf> I don't know - I think they have pretty extensive maths tests though
[23:44:46] <andyf> let me check
[23:45:47] <andyf> HEre - https://github.com/Perl/perl5/blob/blead/ext/POSIX/t/math.t#L23
[23:46:01] <richlowe> did any others fail? :)
[23:46:23] <andyf> I don't know, but I can run them quickly on my box with the > 0 fix
[23:46:49] <richlowe> oh, I thought whoever ran them the first time maybe had results
[23:46:51] <richlowe> but that'd be good
[23:47:15] <andyf> he only mentioned tanh(), but sometimes tests stop after the first failure - I'm not familiar with perl
[23:47:38] <richlowe> and if you want to send that proposed fix to advocates@ for use as a followup, that's probably fine, but it'd be best form-wise if someone other than me pushed it :)
[23:47:44] <richlowe> (and we should fix anything else, too)
[23:48:53] <andyf> tests are running..
[23:55:15] <andyf> ext/POSIX/t/math ............................................... ok
[23:55:59] <igork> andyf: push update :)
[23:56:30] *** merzo <merzo!~merzo@253-53-133-95.pool.ukrtel.net> has quit IRC (Read error: Connection reset by peer)
[23:58:46] <richlowe> ask advocates@ if they're ok with a followup.
[23:59:07] <richlowe> 10881 ... (fix signbit compatibility)
[23:59:08] <richlowe> or something.
[23:59:28] <richlowe> I'll file a bug to fix libm in a second.
top

   June 2, 2019
< | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 30