<html>
<head>
<style>
.hmmessage P
{
margin:0px;
padding:0px
}
body.hmmessage
{
FONT-SIZE: 10pt;
FONT-FAMILY:Tahoma
}
</style>
</head>
<body class='hmmessage'><FONT size=2>
The quake tests on NT386GNU.<BR>
<BR>
m3-sys\m3quake\test<BR>
<BR>
There are the following failures. I can partly explain all of them, and I can get them all to pass, but some of my changes are dubious.<BR>
<BR>
%<BR>
% Executables certainly can be and are writable -- how did they get created in the first place?<BR>
%<BR>
if not equal ($OS, "Windows_NT")<BR>
f = "res = fs_writable( more )"<BR>
check( "f013", f, "" )<BR>
end<BR>
<BR>
so I added:<BR>
<BR>
fs_rmfile("readonly")<BR>
> "readonly" in<BR>
end<BR>
exec("chmod -w readonly")<BR>
f = "res = fs_writable( \"readonly\" )"<BR>
check( "f013b", f, "" )<BR>
<BR>
which works.<BR>
<BR>
f = "res = fs_lsfiles( \"a\", \"T\" )"<BR>
check( "f036", f, "b apple" )<BR>
<BR>
fails because it assumes the file system will return files in a certain order.<BR>
This kind of thing varies legimimately. I added a check2 function that expects<BR>
either of two outputs and it succeeds with:<BR>
<BR>
% some file system sort, some hash, some retain creation order, etc<BR>
% allow either order<BR>
f = "res = fs_lsfiles( \"a\", \"T\" )"<BR>
check2( "f036", f, "b apple", "apple b" )<BR>
<BR>
In this case I think the test is wrong. PPC_DARWIN fails the same here (and otherwise passes; so I can debug)<BR>
<BR>
Tests 8, 9, 10, 11 that verify they can run some pipelines, including sed, do succeed, but they get the wrong output, as if sed didn't run.<BR>
If I wrap the whole thing in sh, it works.<BR>
But that shouldn't be necessary as I understand.<BR>
As I understand q_exec's job is to emulate a bunch of sh behavior, esp. around |, ||, &, &&, >, <, without running sh.<BR>
<BR>
Tests 19 and 20 use q_exec and echo.<BR>
Again a sh wrapper helps, but shouldn't be necessary, as well<BR>
<BR>
echo 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16>x<BR>
yields 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15<BR>
instead of the expected 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16<BR>
If you put a space between the 16 and x, and again the sh wrapper, it works.<BR>
I'd say the 16 is being interpreted as a file handle, but the output does go to the right place, so it's half and half.<BR>
In this case the test might be partly wrong.<BR>
<BR>
Lastly, simply, when you pushd/pop around, you expect BUILD_DIR, not TARGET.<BR>
In this case the test is simply subtely wrong.<BR>
<BR>
echo.exe does exist on Cygwin and it should be what gets run by these.<BR>
echo exists on Darwin (not a shell built in).<BR>
<BR>
I think the ls fix, BUILD_DIR fix, readonly fix, should be taken as I have them, after I test them on PPC_DARWIN.<BR>
The others might be a failure of the Quake code to be consistent across platforms. Someone (maybe me) needs to debug them.<BR>
I also need to test native NT386, but with sed and guess chmod in the path.<BR>
<BR>
- Jay<BR></FONT></body>
</html>