Colm MacCarthaigh wrote:
On Mon, Jul 17, 2006 at 01:11:24PM -0500, William A. Rowe, Jr. wrote:
cd c:\progra~1\Apache~1
so apr_filepath_get returns c:\progra~1\Apache~1
and the TRUENAME resolves to c:\Program Files\Apache Software Foundation.
Do we really want to spend the cycles in
On 7/16/06, William A. Rowe, Jr. [EMAIL PROTECTED] wrote:
To clarify -again-, the case of the name is determined by the case of
the name in the filesystem. The case of the drive is arbitrarily chosen
but must be one or another.
Not true, AFAICT. The MSVC run-time can and does report c:\ or
Justin Erenkrantz wrote:
On 7/16/06, William A. Rowe, Jr. [EMAIL PROTECTED] wrote:
To clarify -again-, the case of the name is determined by the case of
the name in the filesystem. The case of the drive is arbitrarily chosen
but must be one or another.
Not true, AFAICT. The MSVC run-time
On 7/17/06, William A. Rowe, Jr. [EMAIL PROTECTED] wrote:
Comparisons. Deny C:/ pattern. Now, are we going to catch c:/?
If we get TRUECASE of the pattern and the path, they must match if same
or mismatch if different. Case insensitive test is not sufficient since
that's a whopping 96
Justin Erenkrantz wrote:
The problem is that the APR code relies on the MSVC run-time being
consistent: as we have demonstrated, it's not. It can and does report
c:\ in several circumstances.
Yes, and so what? This should be harmless... please indicate the bug
that the VETOED code
On 7/17/06, William A. Rowe, Jr. [EMAIL PROTECTED] wrote:
Justin Erenkrantz wrote:
The problem is that the APR code relies on the MSVC run-time being
consistent: as we have demonstrated, it's not. It can and does report
c:\ in several circumstances.
Yes, and so what? This should be
William A. Rowe, Jr. wrote:
Justin Erenkrantz wrote:
The problem is that the APR code relies on the MSVC run-time being
consistent: as we have demonstrated, it's not. It can and does report
c:\ in several circumstances.
Yes, and so what? This should be harmless... please indicate the bug
Justin Erenkrantz wrote:
On 7/17/06, William A. Rowe, Jr. [EMAIL PROTECTED] wrote:
Justin Erenkrantz wrote:
The problem is that the APR code relies on the MSVC run-time being
consistent: as we have demonstrated, it's not. It can and does report
c:\ in several circumstances.
Yes, and so
Paul Querna wrote:
William A. Rowe, Jr. wrote:
Again, these are the testnames tests that were failing.
Cite them.
I did, in the commit message:
This was causing the test failure on testnames line 219.
Thanks Paul, let's fix that. And if we feel we should TRUENAME the
results of
William A. Rowe, Jr. wrote:
Justin Erenkrantz wrote:
On 7/17/06, William A. Rowe, Jr. [EMAIL PROTECTED] wrote:
Justin Erenkrantz wrote:
The problem is that the APR code relies on the MSVC run-time being
consistent: as we have demonstrated, it's not. It can and does report
c:\ in
On Mon, Jul 17, 2006 at 12:30:09PM -0500, William A. Rowe, Jr. wrote:
Piddle with your experiments in a sandbox. Vetoed code needs to go when
it's vetoed. This veto is over the fact that you've CHANGED security
related behavior, and that won't become acceptable.
The fact that it hasn't
On Mon, Jul 17, 2006 at 11:10:01AM -0700, Justin Erenkrantz wrote:
++1. Something like below? -- justin
I'd test for path[1] == ':', but yep :-)
===
--- filesys.c (revision 421960)
+++ filesys.c (working copy)
@@ -193,6
Colm MacCarthaigh wrote:
maybe there's a case for modifying apr_filepath_get on win32 here, to
use the same canonicalised format? We're probably not the only people
likely to perform these kind of niaive string comparisons, and we're
supposed to know our own codebase ;-)
Very true :)
Paul Querna wrote:
Er. I'm sorry I didn't revert it in under 24 hours, but like, I wasn't
just reading email and hanging out with my machine ready to revert a
commit right after I made it. Like i said earlier, I will try to get to
it tonight, we all have busy schedules, so don't lay down this
On Mon, Jul 17, 2006 at 01:11:24PM -0500, William A. Rowe, Jr. wrote:
cd c:\progra~1\Apache~1
so apr_filepath_get returns c:\progra~1\Apache~1
and the TRUENAME resolves to c:\Program Files\Apache Software Foundation.
Do we really want to spend the cycles in filepath_get to normalize this
Colm MacCarthaigh wrote:
On Mon, Jul 17, 2006 at 01:11:24PM -0500, William A. Rowe, Jr. wrote:
cd c:\progra~1\Apache~1
so apr_filepath_get returns c:\progra~1\Apache~1
and the TRUENAME resolves to c:\Program Files\Apache Software Foundation.
Do we really want to spend the cycles in
Damnit, you really want more case-canonical flaws in httpd?
I don't give a shit if it's force-lower or force-upper. It so happens
that 90% of the time, the NT Kernel represents drive letters in upper
case. But it's neither here nor there, revert this.
Oh - btw - nice job Justin and Paul on
William A. Rowe, Jr. wrote:
Damnit, you really want more case-canonical flaws in httpd?
So, is that a -1 or just a complaint?
I don't give a shit if it's force-lower or force-upper. It so happens
that 90% of the time, the NT Kernel represents drive letters in upper
case. But it's neither
Paul Querna wrote:
William A. Rowe, Jr. wrote:
Damnit, you really want more case-canonical flaws in httpd?
So, is that a -1 or just a complaint?
That part was a question after my less-than-fun trawl through svn blame
on server/request.c
I don't give a shit if it's force-lower or
19 matches
Mail list logo