Re: [9fans] Compiler Message

2014-07-21 Thread tlaronde
On Mon, Jul 21, 2014 at 12:55:44PM +1000, Shane Morris wrote:

 clang: error: unknown argument: '-mno-fused-madd'
 [-Wunused-command-line-argument-hard-error-in-future]
 clang: note: this will be a hard error (cannot be downgraded to a warning)
 in the future

Apparently the future hard error is treated as an error.

It looks like clang does not understand the gcc argument
'-mno-fused-madd'. Only the developers of Inferno would know if
multiply-add for float caveat is really a requisite for Inferno.
So look if there is the equivalent in clang, and convert the
argument; or suppress it where it is defined in the mkfiles (perhaps
it is necessary for gcc, doing things, and not necessary with
another compiler, so the compiled program will not exhibit any
problem).

HTH
-- 
Thierry Laronde tlaronde +AT+ polynum +dot+ com
  http://www.kergis.com/
  http://www.renaissance-francaise.fr/
Key fingerprint = 0FF7 E906 FBAF FE95 FD89  250D 52B1 AE95 6006 F40C



Re: [9fans] Compiler Message

2014-07-21 Thread Ramakrishnan Muthukrishnan
On Mon, Jul 21, 2014 at 8:25 AM, Shane Morris edgecombe...@gmail.com wrote:
 Hello again 9fans,

 I'm also trying to compile hosted Inferno for OS X 10.9, all seems to go
 well until the mk install giving this error message:

 shanes-air-2:inferno-os boris$ PATH=`pwd`/MacOSX/386/bin:$PATH mk install
 (cd lib9; mk  install)
 cc -c -arch i386 -mmacosx-version-min=10.4 -Wno-deprecated-declarations
 -Wuninitialized -Wunused -Wreturn-type -Wimplicit -Wno-four-char-constants
 -Wno-unknown-pragmas -pipe -fno-strict-aliasing -no-cpp-precomp
 -mno-fused-madd -I/Users/boris/Documents/inferno-os/MacOSX/386/include
 -I/Users/boris/Documents/inferno-os/include -Os convD2M.c
 clang: error: unknown argument: '-mno-fused-madd'
 [-Wunused-command-line-argument-hard-error-in-future]
 clang: note: this will be a hard error (cannot be downgraded to a warning)
 in the future
 mk: cc -c -arch ...  : exit status=exit(1)
 mk: for j in ...  : exit status=exit(1)
 shanes-air-2:inferno-os boris$

 Does anyone have any insight?

Hi,

On OS X 10.9.x, gcc points to clang. I installed gcc-4.9.0 from source
(follow instructions on this page, for example:
http://solarianprogrammer.com/2013/06/11/compiling-gcc-mac-os-x/ to
get a working gcc. I don't use a package manager on OSX these days, I
compile/install what I need, from source.)

I then edited mkfiles/mkfile-MacOSX-386 to point to the newly built
gcc (I called the gcc binary gcc-4.9.0). Here is the complete file.

TARGMODEL=  Posix
TARGSHTYPE= sh
CPUS=   386

O=  o
OS= o

AR= ar
ARFLAGS=ruvs
A=  a

AS= gcc-4.9.0 -c -arch i386 -m32
ASFLAGS=

ISYSROOT=   -isysroot /Developer/SDKs/MacOSX10.6.sdk

CC= gcc-4.9.0 -c -m32
COPTFLAGS=  -Os
CDEBUGFLAGS=
CTHREADFLAGS=
CFLAGS= -arch i386 -m32\
-mmacosx-version-min=10.6\
-Wno-deprecated-declarations -Wuninitialized -Wunused
-Wreturn-type -Wimplicit -Wno-four-char-constants
-Wno-unknown-pragmas\
-pipe\
-fno-strict-aliasing\
-mno-fused-madd\
-I$ROOT/MacOSX/386/include\
-I$ROOT/include\
$COPTFLAGS $CDEBUGFLAGS\

LD= gcc -arch i386 -m32
LDFLAGS=\
-mmacosx-version-min=10.4\
-multiply_defined suppress

SYSLIBS=

YACC=   iyacc
YFLAGS= -d


-- 
  Ramakrishnan



Re: [9fans] Compiler Message

2014-07-21 Thread dante

Hi Shane,

Removing the compile argument won't help (leads to errors somewhere 
else).

I think that the most expedient solution is Ramkrishnan's.

In order to catch this sort of errors, some sort of continuous 
integration would be needed.
I have no idea how this could be done without investing into a real Mac 
machine, which is costly.


I reported the error to the person shown in the Hg commits; there was 
unfortunately no response yet.
Does anyone know which is the proper channel to report issues with 
Inferno?


Shane, I'm just curious: did you try to install Inferno for reasons 
similar to mine (here below)?


In tried to install Inferno because I thought that it was the simplest 
way to access (at least the file system of) my Plan9 Raspberry Pi from 
my Mac.

The other solutions I tried are awkward:
1. Plan9 under Virtual Box (don't want to pay for VMWare/Parallels).
Configured network card according to Richard Miller's suggestion.
Boots, but sometimes it doesn't, requiring to restart the program.
Network works only sometimes.
2. Plan9 from user space, installed over Brew.
It works, but I don't see there any way to import (mount) the 
remote file system locally.


Cheers,
Dante

On 21.07.2014 04:55, Shane Morris wrote:

Hello again 9fans,

I'm also trying to compile hosted Inferno for OS X 10.9, all seems to
go well until the mk install giving this error message:

 shanes-air-2:inferno-os boris$ PATH=`pwd`/MacOSX/386/bin:$PATH mk 
install

(cd lib9; mk  install)
cc -c -arch i386 -mmacosx-version-min=10.4
-Wno-deprecated-declarations -Wuninitialized -Wunused -Wreturn-type
-Wimplicit -Wno-four-char-constants -Wno-unknown-pragmas -pipe
-fno-strict-aliasing -no-cpp-precomp -mno-fused-madd
-I/Users/boris/Documents/inferno-os/MacOSX/386/include
-I/Users/boris/Documents/inferno-os/include -Os convD2M.c
clang: error: unknown argument: '-mno-fused-madd'
[-Wunused-command-line-argument-hard-error-in-future]
clang: note: this will be a hard error (cannot be downgraded to a
warning) in the future
mk: cc -c -arch ...  : exit status=exit(1)
mk: for j in ...  : exit status=exit(1)
shanes-air-2:inferno-os boris$

Does anyone have any insight?

Many thanks!

Shane.




Re: [9fans] Compiler Message

2014-07-21 Thread Shane Morris
No, the reasons I attempted an install is because I have no experience
getting hosted Inferno to run in Plan 9, although I possess a couple of
stable Plan 9 VMs (when they can reach the network, they'll ping 8.8.8.8,
but not get any sites like sources or 9gridchan - to be fair, there seems
to be network problems on those sites at the exact times I want to do
something, as I can't mount sources from Mac9P either), and I wish to have
an ARM Inferno compiler to start compiling against the Parallella,
following the guide at lynxline blog, and the boot up process at
parallella.org. I am yet to try anything, I think I'll leave it to tomorrow
at this stage.


On Mon, Jul 21, 2014 at 6:12 PM, dante subscripti...@posteo.eu wrote:

 Hi Shane,

 Removing the compile argument won't help (leads to errors somewhere else).
 I think that the most expedient solution is Ramkrishnan's.

 In order to catch this sort of errors, some sort of continuous
 integration would be needed.
 I have no idea how this could be done without investing into a real Mac
 machine, which is costly.

 I reported the error to the person shown in the Hg commits; there was
 unfortunately no response yet.
 Does anyone know which is the proper channel to report issues with Inferno?

 Shane, I'm just curious: did you try to install Inferno for reasons
 similar to mine (here below)?

 In tried to install Inferno because I thought that it was the simplest way
 to access (at least the file system of) my Plan9 Raspberry Pi from my Mac.
 The other solutions I tried are awkward:
 1. Plan9 under Virtual Box (don't want to pay for VMWare/Parallels).
 Configured network card according to Richard Miller's suggestion.
 Boots, but sometimes it doesn't, requiring to restart the program.
 Network works only sometimes.
 2. Plan9 from user space, installed over Brew.
 It works, but I don't see there any way to import (mount) the remote
 file system locally.

 Cheers,
 Dante


 On 21.07.2014 04:55, Shane Morris wrote:

 Hello again 9fans,

 I'm also trying to compile hosted Inferno for OS X 10.9, all seems to
 go well until the mk install giving this error message:

  shanes-air-2:inferno-os boris$ PATH=`pwd`/MacOSX/386/bin:$PATH mk
 install
 (cd lib9; mk  install)
 cc -c -arch i386 -mmacosx-version-min=10.4
 -Wno-deprecated-declarations -Wuninitialized -Wunused -Wreturn-type
 -Wimplicit -Wno-four-char-constants -Wno-unknown-pragmas -pipe
 -fno-strict-aliasing -no-cpp-precomp -mno-fused-madd
 -I/Users/boris/Documents/inferno-os/MacOSX/386/include
 -I/Users/boris/Documents/inferno-os/include -Os convD2M.c
 clang: error: unknown argument: '-mno-fused-madd'
 [-Wunused-command-line-argument-hard-error-in-future]
 clang: note: this will be a hard error (cannot be downgraded to a
 warning) in the future
 mk: cc -c -arch ...  : exit status=exit(1)
 mk: for j in ...  : exit status=exit(1)
 shanes-air-2:inferno-os boris$

 Does anyone have any insight?

 Many thanks!

 Shane.





Re: [9fans] Compiler Message

2014-07-21 Thread cam
 In tried to install Inferno because I thought that it was the simplest 
 way to access (at least the file system of) my Plan9 Raspberry Pi from 
 my Mac.
 The other solutions I tried are awkward:
 1. Plan9 under Virtual Box (don't want to pay for VMWare/Parallels).
  Configured network card according to Richard Miller's suggestion.
  Boots, but sometimes it doesn't, requiring to restart the program.
  Network works only sometimes.

have you tried qemu?  plan9 works wonderfully as a guest.
network access is as reliable as your mac.

 2. Plan9 from user space, installed over Brew.
  It works, but I don't see there any way to import (mount) the 
 remote file system locally

install macfuse and then you can use 9pfuse.  it works, though
it is a bit slow.




Re: [9fans] Compiler Message

2014-07-21 Thread Shane Morris
Q for Mac is horrible, dodgy software. Then again, I didn't build it from
source. The good thing is, it'll convert images just fine.


On Mon, Jul 21, 2014 at 6:39 PM, c...@9.squish.org wrote:

  In tried to install Inferno because I thought that it was the simplest
  way to access (at least the file system of) my Plan9 Raspberry Pi from
  my Mac.
  The other solutions I tried are awkward:
  1. Plan9 under Virtual Box (don't want to pay for VMWare/Parallels).
   Configured network card according to Richard Miller's suggestion.
   Boots, but sometimes it doesn't, requiring to restart the program.
   Network works only sometimes.

 have you tried qemu?  plan9 works wonderfully as a guest.
 network access is as reliable as your mac.

  2. Plan9 from user space, installed over Brew.
   It works, but I don't see there any way to import (mount) the
  remote file system locally

 install macfuse and then you can use 9pfuse.  it works, though
 it is a bit slow.





Re: [9fans] Compiler Message

2014-07-21 Thread dante

What about Drawterm on Mac? Is it working well?
This would require of course configuring a CPU server...

Thanks,
Dante

On 21.07.2014 10:39, c...@9.squish.org wrote:
In tried to install Inferno because I thought that it was the 
simplest
way to access (at least the file system of) my Plan9 Raspberry Pi 
from

my Mac.
The other solutions I tried are awkward:
1. Plan9 under Virtual Box (don't want to pay for VMWare/Parallels).
 Configured network card according to Richard Miller's 
suggestion.
 Boots, but sometimes it doesn't, requiring to restart the 
program.

 Network works only sometimes.


have you tried qemu?  plan9 works wonderfully as a guest.
network access is as reliable as your mac.


Will do.




2. Plan9 from user space, installed over Brew.
 It works, but I don't see there any way to import (mount) the
remote file system locally


install macfuse and then you can use 9pfuse.  it works, though
it is a bit slow.




Re: [9fans] Compiler Message

2014-07-21 Thread Shane Morris
Using the ANTS CPU server image, and Drawterm-Cocoa.


On Mon, Jul 21, 2014 at 6:43 PM, dante subscripti...@posteo.eu wrote:

 What about Drawterm on Mac? Is it working well?
 This would require of course configuring a CPU server...

 Thanks,
 Dante


 On 21.07.2014 10:39, c...@9.squish.org wrote:

 In tried to install Inferno because I thought that it was the simplest
 way to access (at least the file system of) my Plan9 Raspberry Pi from
 my Mac.
 The other solutions I tried are awkward:
 1. Plan9 under Virtual Box (don't want to pay for VMWare/Parallels).
  Configured network card according to Richard Miller's suggestion.
  Boots, but sometimes it doesn't, requiring to restart the program.
  Network works only sometimes.


 have you tried qemu?  plan9 works wonderfully as a guest.
 network access is as reliable as your mac.


 Will do.



  2. Plan9 from user space, installed over Brew.
  It works, but I don't see there any way to import (mount) the
 remote file system locally


 install macfuse and then you can use 9pfuse.  it works, though
 it is a bit slow.





Re: [9fans] Compiler Message

2014-07-21 Thread cam
 Q for Mac is horrible, dodgy software. Then again, I didn't build it from
 source. The good thing is, it'll convert images just fine.

older versions probably.  i built qemu-1.7.0 from source before
2.0 was stable and have had no problems.  i own fusion, but i 
prefer qemu since fusion occupies kernel memory (meaning an 
idle vm cannot be swapped out), and it uses about 30% more 
memory per vm than qemu does.  

what problems did you have with it?




Re: [9fans] Compiler Message

2014-07-21 Thread Shane Morris
The older versions refused to run. I've not tried this version I downloaded
today, only the qemu-img inside the package which worked fine.


On Mon, Jul 21, 2014 at 6:55 PM, c...@9.squish.org wrote:

  Q for Mac is horrible, dodgy software. Then again, I didn't build it from
  source. The good thing is, it'll convert images just fine.

 older versions probably.  i built qemu-1.7.0 from source before
 2.0 was stable and have had no problems.  i own fusion, but i
 prefer qemu since fusion occupies kernel memory (meaning an
 idle vm cannot be swapped out), and it uses about 30% more
 memory per vm than qemu does.

 what problems did you have with it?





Re: [9fans] Compiler Message

2014-07-21 Thread cam
 What about Drawterm on Mac? Is it working well?

yes.  i use drawterm from 

http://code.swtch.com/drawterm

i compiled it to run under X since i have to run that beast 
anyways, but there is also drawterm-cocoa which builds 
and runs native under osx.

http://bitbucket.org/jas/drawterm-cocoa

 This would require of course configuring a CPU server...

if you want the full experience of plan9, that is a given.  it
is not hard to do; and once you have a cpu/auth server,
booting additional cpu servers is an incredibly trivial task.




Re: [9fans] raspberry pi model B+

2014-07-21 Thread kokamoto
 the B+ has some power supply improvements and two more USB
 ports, which are said to make it work better with usb hard drives.

You mean B+ may be able to use as a Plan9 file server?

Kenji




[9fans] The Third Button

2014-07-21 Thread dante

Dear 9fans,

Is there any situation (i.e., GUI area) where both the 2nd *and* the 
3rd mouse button are indispensable options?

Except for chording.

Would it be possible to create the option of merging these two buttons 
for machines not blessed with the traditional rodent?


I ask this because 3-button-mice (*not* 2-button-and-wheel) are almost 
impossible to find nowadays.
Moreover, many notebooks have good touchpads to be used in situations 
when mice are impractical (on train, on the lap).
These interfaces present pretty nice solutions for button-1 and 
button-3, but it's hard to impossible to do a button-2 click.


In Acme, for instance, you generally need button-3 in the text area and 
button-2 in the menu area.
I'm not sure when button-2 in the text area and button-3 in the menu 
area would be indispensable.
In Rio, you generally need button-3 when you click outside the focused 
window and button-2 when you click inside the focused window.


Cheers,
Dante




Re: [9fans] The Third Button

2014-07-21 Thread Robert Raschke
In acme, button-3 can cancel a button-2 execute. Say you button-2-sweep a
command, but then decide, err, no, don't want to do that, you chord-click
button-3 to cancel the execution. Not sure about other ways of doing this.


Robby


On 21 July 2014 14:08, dante subscripti...@posteo.eu wrote:

 Dear 9fans,

 Is there any situation (i.e., GUI area) where both the 2nd *and* the 3rd
 mouse button are indispensable options?
 Except for chording.

 Would it be possible to create the option of merging these two buttons for
 machines not blessed with the traditional rodent?

 I ask this because 3-button-mice (*not* 2-button-and-wheel) are almost
 impossible to find nowadays.
 Moreover, many notebooks have good touchpads to be used in situations when
 mice are impractical (on train, on the lap).
 These interfaces present pretty nice solutions for button-1 and button-3,
 but it's hard to impossible to do a button-2 click.

 In Acme, for instance, you generally need button-3 in the text area and
 button-2 in the menu area.
 I'm not sure when button-2 in the text area and button-3 in the menu area
 would be indispensable.
 In Rio, you generally need button-3 when you click outside the focused
 window and button-2 when you click inside the focused window.

 Cheers,
 Dante





Re: [9fans] The Third Button

2014-07-21 Thread cam
with drawterm on osx, button 2 click is option+click, button 3
is command+click.  

in qemu, button 2 is shift+command+click and button 3 is
command+click.

on pc's with ps/2 two button mice, i believe shift+click is 
button 2.

if you are using something different, try various combinations
of modifier keys and clicking.  




Re: [9fans] The Third Button

2014-07-21 Thread cam
i forgot to mention: alternately, kbmap(3) is supposed to
let you use function keys as mouse buttons.




Re: [9fans] raspberry pi model B+

2014-07-21 Thread Richard Miller
 the B+ has some power supply improvements and two more USB
 ports, which are said to make it work better with usb hard drives.
 
 You mean B+ may be able to use as a Plan9 file server?

It may be more reliable but I would still expect it to be slow.  I don't
have a usb hard drive so I can't give any experimental results.




[9fans] interleaved output of ratrace

2014-07-21 Thread Nick Owens
hello,

i'm trying to use ratrace to debug a go binary.
what i don't understand is why records of syscalls sometimes appear on
top of one another.

here's the output of 'ratrace -c ./strings.test [2]/tmp/ratrace.log',
where strings.test is made with 'go test -c strings'.

http://sprunge.us/aCBG

you can see in several places that more than one syscall record is
printed on the same line.

i find this quite odd since in /sys/src/cmd/ratrace.c:/^writer all
output comes from a channel.

i'm running this on 9front/amd64, but i'd appreciate if anyone can test
on other systems.



pgpGVESQjld0m.pgp
Description: PGP signature