Checked in a Rakefile that should work correctly with Linux filesystem

Can someone with a Linux box please verify that this works correctly?

Thanks,
-John

Sanghyeon S.:

:compile task depends on :happy task, which cannot find .exe commands.

Excellent! I just applied this patch to my local copy. I’ll push it out
tomorrow along with some other fixes that I’m working on tonight.

Other than that, does it compile and run OK on Mono?

Thanks,
-John

2007/10/25, John L. (DLR) [email protected]:

Excellent! I just applied this patch to my local copy. I’ll push it out tomorrow along with some other fixes that I’m working on tonight.

Other than that, does it compile and run OK on Mono?

COMMON_DEBUG_SWITCHES has define:DEBUG;TRACE, but the semicolon is a
statement separator in *nix shell. I think you need to quote it.

Other than that, Mono 1.2.5 (the latest stable release) will suffer
from:
https://bugzilla.novell.com/show_bug.cgi?id=325110
https://bugzilla.novell.com/show_bug.cgi?id=325319
https://bugzilla.novell.com/show_bug.cgi?id=333647

Thankfully, all of them are fixed in current SVN. IronRuby compiles
and runs OK where OK is defined as evaluating 1+1. I haven’t tried
tests yet.

2007/10/25, John L. (DLR) [email protected]:

Can someone with a Linux box please verify that this works correctly?

:compile task depends on :happy task, which cannot find .exe commands.
The following patch should correct that.

Index: Rakefile

— Rakefile (revision 48)
+++ Rakefile (working copy)
@@ -597,10 +597,10 @@
puts “***** run the script from within a MERLIN command prompt.”
end

  • commands = [‘resgen.exe’, ‘csc.exe’]
  • commands = ENV[‘mono’].nil? ? [‘resgen.exe’, ‘csc.exe’] : [‘resgen’,
    ‘gmcs’]
    commands += [‘tf.exe’, ‘svn.exe’] if MERLIN_ROOT
  • paths = ENV[‘PATH’].split(‘;’).collect { |path| Pathname.new path }
  • paths = ENV[‘PATH’].split(File::PATH_SEPARATOR).collect { |path|
    Pathname.new path }

    failure = false
    commands.each do |command|

2007/10/26, John L. (DLR) [email protected]:

I tried building using a binary redist of Mono for the Mac, and ran into these bugs. I then switched over to the Mono SVN repository and saw a regression - it fails to compile at all with a new compiler error (see below). I just sent an email to Miguel to see if he could look into this for us rather than us trying to figure out what’s causing gmcs to fail.

Interesting that this worked for you though …

Yes, it was a very recent regression, and it is now fixed. It did work
sometime between 1.2.5 and SVN of yesterday…

I updated to SVN r88240 and it’s all fine now.

Sanghyeon S.:

COMMON_DEBUG_SWITCHES has define:DEBUG;TRACE, but the semicolon is a
statement separator in *nix shell. I think you need to quote it.

I’m trying to build IronRuby on my Mac now. Quoting does fix the problem
above.

Other than that, Mono 1.2.5 (the latest stable release) will suffer
from:
Access Denied
Access Denied
Access Denied

I tried building using a binary redist of Mono for the Mac, and ran into
these bugs. I then switched over to the Mono SVN repository and saw a
regression - it fails to compile at all with a new compiler error (see
below). I just sent an email to Miguel to see if he could look into this
for us rather than us trying to figure out what’s causing gmcs to fail.

Thankfully, all of them are fixed in current SVN. IronRuby compiles and
runs OK where OK is defined as evaluating 1+1. I haven’t tried tests
yet.

Interesting that this worked for you though …

Thanks,
-John

/Users/jflam/dev/ironruby/src/microsoft.scripting/RuntimeHelpers.cs(266,50):
error CS0111: A member
Microsoft.Scripting.RuntimeHelpers.CreateSimpleCallSite<T0,T1,R>(Microsoft.Scripting.CodeContext)' is already defined. Rename this member or use different parameter types /Users/jflam/dev/ironruby/src/microsoft.scripting/RuntimeHelpers.cs(262,46): (Location of the symbol related to previous error) /Users/jflam/dev/ironruby/src/microsoft.scripting/RuntimeHelpers.cs(270,54): error CS0111: A member Microsoft.Scripting.RuntimeHelpers.CreateSimpleCallSite<T0,T1,T2,R>(Microsoft.Scripting.CodeContext)’
is already defined. Rename this member or use different parameter types
/Users/jflam/dev/ironruby/src/microsoft.scripting/RuntimeHelpers.cs(262,46):
(Location of the symbol related to previous error)
/Users/jflam/dev/ironruby/src/microsoft.scripting/RuntimeHelpers.cs(274,58):
error CS0111: A member
Microsoft.Scripting.RuntimeHelpers.CreateSimpleCallSite<T0,T1,T2,T3,R>(Microsoft.Scripting.CodeContext)' is already defined. Rename this member or use different parameter types /Users/jflam/dev/ironruby/src/microsoft.scripting/RuntimeHelpers.cs(262,46): (Location of the symbol related to previous error) /Users/jflam/dev/ironruby/src/microsoft.scripting/RuntimeHelpers.cs(278,62): error CS0111: A member Microsoft.Scripting.RuntimeHelpers.CreateSimpleCallSite<T0,T1,T2,T3,T4,R>(Microsoft.Scripting.CodeContext)’
is already defined. Rename this member or use different parameter types
/Users/jflam/dev/ironruby/src/microsoft.scripting/RuntimeHelpers.cs(262,46):
(Location of the symbol related to previous error) Compilation failed: 4
error(s), 0 warnings

Sanghyeon S.:

Interesting that this worked for you though …

Yes, it was a very recent regression, and it is now fixed. It did work
sometime between 1.2.5 and SVN of yesterday…

I updated to SVN r88240 and it’s all fine now.

That’s great news! I’ll update via SVN when I get home (I can’t hit
svn:// from inside of MSFT on my Mac since we don’t have a proxy client
for Mac handy).

BTW, have you tried running the tests yet? You could try the simpler
ones via rake test, or the more adventurous ones via rake specs.

-John