I've seen this behaviour too with my 1.1.3 build for s390. (It works on one
system but not another and I can't work out why.)
-Original Message-
> You probably did this before and left a /tmp/mod_mono_server and
> a /tmp/.wapi directory owned by the root user. Remove that file and the
> d
I'll attempt to think before I type this time :-)
I have a a C routine I wish to call that takes the following parameters:
typedef struct XXXCB {
short a;
char b[2];
};
void scumbag(XXXCB *cb, char *v);
for our purposes v is an arbitrarily long string of bytes in the range 0x00-0x7f.
I've implemented some custom marshalling for an external call so that the parameters
being sent gets massaged as required by the external routine. In addition, it is
supposed to massage data after the call. However, while it will do so on the way out,
it's not being driven the other way. I found
int scumbag(int *X, char *fb, char *rb ...)
(the int values were the for the marshalling code that uses the sizeparamindex value
to determine the size of the byte arrays)
-Original Message-
You should not use ByValArray enumeration member on function arguments. MS.NET
SDK documentation
Thanks Dmitry,
I tried this and am getting the following when compiling this simple test case:
using System;
using System.Text;
using System.Runtime.InteropServices;
public class main {
[DllImport("libname", EntryPoint = "scumbag")]
private unsafe static extern i
I have a variable:
byte[] var;
That I wish to pass to a C program. Normally, I could use something like:
[MarshalAs(UnmanagedType.ByValArray, SizeConst=8)]
byte[] var;
But in my case I do not know the size of the variable until runtime. How do I get
something like this marshalled at runtime?
Ne
I am trying to write a .Net interface to a system of ours. The system takes the
following structure:
struct XXXid {
short level;
short size;
charnode[8];
char user[8];
int no;
}
Both user & node a 8 byte character strings padded with blanks and no
When building a test app I received:
error CS0233: Sizeof may only be used in an unsafe context (consider using
System.Runtime.InteropServices.Marshal.Sizeof
However, the actual method is ...SizeOf.
Neale Ferguson
___
Mono-list maillist - [EMAIL PR
Hi,
I'm playing
with TestService.asmx and have used wsdl to build the .cs file and have complied
it into a library. What I want to know is where do I put the resulting DLL so
that the server can use it and stop reporting:
Compilation ErrorDescription:
Error compiling a resource required to
Running wsdl://locahost/demo/TestService.asmx I see the following in access_log:
127.0.0.1 - - [01/Oct/2004:12:25:35 -0400] "GET /demo/TestService.asmx HTTP/1.1" 200
3022 "-" "Mono Web Services Client Protocol 1.1.4322.573"
127.0.0.1 - - [01/Oct/2004:12:25:40 -0400] "GET /demo/demo/TestService.as
I just refreshed with monocharge-latest (and then 20040927) and my builds of mcs are
resulting in pages of complaints like:
modifiers.cs(8) error CS8025: Parsing error
syntax error, got token `CLASS', expecting ABSTRACT AS EXTERN IN INTERNAL IS NEW
OPERATOR OVERRIDE PRIVATE PROTECTED PUBLIC READ
Is there anything in /var/log/[apache2|apache|http]/error_log?
-Original Message-
Golo Haas wrote:
>Hi there,
>
>I have just installed mod_mono 1.0.1 and I am trying to get it to run
with Mono 1.0.1 on Novell SuSE Linux 9.1, using Apache 2.0.49.
>
>The problem I have is that mod_mono does
I fixed it (and the one in net_2_0) with the following:
--- runtime/net_1_1/Makefile.am 7 Sep 2004 22:08:33 - 1.46
+++ runtime/net_1_1/Makefile.am 16 Sep 2004 02:17:47 -
@@ -80,11 +80,11 @@
corlib_DATA = mscorlib.dll
EXTRA_DIST = $(gac_assemblies) $(corlib_DATA) $(mdb_files)
MAIN
After fixing the Makefile.am problem described in an earlier post, I'm now getting the
following and don't know how to proceed/fix:
make[1]: Entering directory `/home/usanefe/Mono/mono/runtime'
list='. net_1_1 net_2_0'; for subdir in $list; do \
if test "$subdir" = .; then :; else \
test -d
Hi,
I had problems building mono cleanly owing to missing items in runtime/Makefile.am.
The following patch fixes things. Before I commit I wanted to run it by the group...
--- Makefile.am 9 Sep 2004 20:29:55 - 1.83
+++ Makefile.am 13 Sep 2004 13:53:23 -
@@ -97,10 +97,17 @@
I found the problem. It was due to an update to libgc/configure.in. There is a test
for linux so as to set a flag. S/390 was missing from the test so this didn't get set
and so the GC_local_malloc function was not being compiled.
--
Af
After a very pleasnt 3 weeks in Australia I returned and did a cvs update; make clean;
make. The resulting build gives me:
/home/usanefe/Mono/mono/mono/metadata/domain.c:262: undefined reference to
`GC_local_malloc'
../utils/.libs/libmonoutils.a(mono-hash.o)(.text+0x824): In function
`g_hash_ta
I added the flag and my TAILC stuff works perfectly! The only remaining item (for
32-bit) is implementing OP_ARGLIST. From reading the doc. and the other platform's
code I still am not sure what it's supposed to do.
-Original Message-
Just add a new flag to MonoCompileFlags in mini.h and
For environments like S/390 the use of X will be quite limited: most of the work will
be web services type stuff. A "frill-free" package would be great .
Neale
___
Mono-list maillist - [EMAIL PROTECTED]
http://lists.ximian.com/mailman/listinfo/mono-li
I just rebuilt my JIT using the latest monocharge dated July 15 and am getting:
Corlib not in sync with this runtime: expected corlib version 23, found 22.
Download a newer corlib or a newer runtime at http://www.go-mono.com/daily.
Is there an update scheduled for the daily site?
Neale Ferguson
Hi,
I'm porting the JIT to 64-bit S/390. The 32-bit version uses two registers for long
type variables. The 64-bit only uses one. I modified the code so that a long value
returned from a routine is copied from the ABI defined register to the destination
register (previously it would copy from t
The link (http://go-mono.com/daily) from the compilation page
(http://mono-project.com/contributing/compiling.html) is timing out when I attempt to
connect.
Neale Ferguson
___
Mono-list maillist - [EMAIL PROTECTED]
http://lists.ximian.com/mailman/l
Bad update of what? This is what I get:
98: [PEAPI]PEAPI.CustomModifier
If I do a cvs update -Pd . and rebuild will that do it?
-Original Message-
That's probably the result of a bad update. Anyway, if you want to know
what's missing, run 'monodis --typeref ilasm.exe | grep ^98' and you'll
The S/390 ABI only specifies that Registers 6-14 are guaranteed on exit as they were
on entry. It does provide a place on the stack for parameter registers to be place
there but for performance reasons it's better not to. If there was a flag set for the
method that the emit_prolog could check to
It's the one found in the tests subdirectory. I was running the test
suite after rebuidling mono.
-Original Message-
Do you have the "vararg.il" available for us to look at? The il
assembler
is probably legitimately complaining how there is no constructor for an
object.
__
I'm not sure what this means, can someone point me at what may need fixing?
ilasm -out:vararg.exe vararg.il
Assembling 'vararg.il' , no listing file, to exe --> 'vararg.exe'
** (/opt/mono/bin/ilasm.exe:5586): WARNING **: Missing method .ctor in assembly
/opt/mono/bin/ilasm.exe typeref index 98
BTW: The only reason I'm bothering with this is because the mono build process craps
out when it tests for the existence of ../../gmcs/gmcs.exe.
-Original Message-
I did but the make process (from the top directory) doesn't even enter
that directory to do its thing. When I go into the dir
I did but the make process (from the top directory) doesn't even enter
that directory to do its thing. When I go into the directory and run the
make I get the junk below..
-Original Message-
On Fri, 2004-07-23 at 12:30, [EMAIL PROTECTED] wrote:
> Is this "normal"?
>
> make[1]: Entering di
Is this "normal"?
make[1]: Entering directory `/home/usanefe/mcs/gmcs'
MONO_PATH="../class/lib/net_2_0_bootstrap:$MONO_PATH" mono ../mcs/mcs.exe -2
/target:exe /debug /out:gmcs.exe AssemblyInfo.cs anonymous.cs assign.cs attribute.cs
driver.cs cs-tokenizer.cs cfold.cs class.cs codegen.cs const.c
-Original Message-
Run it through apache with your configuration and get the command line
from 'ps -aux'. Kill mod-mono-server. Then run that command line as the
same user that runs apache from a console.
- I do but because it makes no socket connection I don't get the same behavior.
You
I should add running
mono xsp.exe works
fine.
I've built and
installed mod_mono for my apache2 server. When I try and get to it I'd get
Server Unavailable errors and a message in error_log:
** ERROR **: file exceptions-s390.c: line 837
(ves_icall_get_trace): assertion failed: (ji !=
NULL)aborting...
I added a
syslog call to the S/3
Hi,
The implementation of tailc is quite x86-centric in mini.c. It assumes that all
parameters are on the stack and by simply going back to the previous stack frame then
the jump to the new routine will have access to the parameters that it wants. For
architectures where parameters can be in re
Hi,
Well I might
be passing all but two regression tests with my S390 JIT now, but when I try and
use it to build mcs I don't get passed the first compile!!
Unhandled Exception: System.ArgumentOutOfRangeException: Index is less
than 0 or more than or equal to the list count.Parameter name:
There's a couple of idiosycracies with S390.
For example, in the exception8 test:using System;public class TryTest
{ public static void
ThrowException()
{
throw new Exception();
} public static int Main()
{
int state =
0;
Hi,
I am now passing all but two of the regression tests using my JIT port. One is
strange that I haven't fully diagnosed with appdomain-unload (the page containing one
of the compiled methods seems to get freed); the other is relates to single floating
point numbers.
The way I handle single f
Hi again,
The question of exception handling has just arised as I run some more tests for
S/390. For example, if I handle and exception and need to call a filter where I have
to restore the stack pointer to the frame where the "finally" exists. If this bit of
code then issues its own calls then
Title: Endian issue
My S390 JIT is almost doing the right thing. Apart from some exception handling issues things look pretty good. However, when I run the compiler with the JIT the byte ordering is a problem. For example I produce:
000 4d5a 0600 0400
instead of
I am getting the s390 JIT fixed for the beta 1 level of code. I'm testing the
exception handling and my test case hangs as the backtrace information (I believe) is
being created. There's a call for a relatively large amount of storage which gets the
gc stuff involved. The whole process just hang
39 matches
Mail list logo