On Thu, Oct 17, 2013 at 11:08 PM, Gregg Smith <gls@gknw.net> wrote:
On 10/17/2013 7:49 AM, Jeff Trawick wrote:
On Wed, Oct 16, 2013 at 9:22 AM, Jeff Trawick <trawick@gmail.com <mailto:trawick@gmail.com>> wrote:

    On Tue, Oct 1, 2013 at 8:26 AM, Jeff Trawick <trawick@gmail.com
    <mailto:trawick@gmail.com>> wrote:

        On Tue, Oct 1, 2013 at 7:32 AM, Jim Jagielski <jim@jagunet.com
        <mailto:jim@jagunet.com>> wrote:

            Subj sez it all.

            I've never been RM for APR but am willing to do this one...


        I'm happy to take care of the RM work.

        Some things I'm interested in checking on for 1.5.x:

        * Compare 1.4.x<->1.5.x and see if anything was missed.  (I'm
        pretty sure that there are some .h fixes that skipped 1.5.x;
        dunno what else.)
        * Add the cmake support.  (I purposefully skipped the 1.5.x
        branch when there was constant activity.)
        * Put in a couple of fixes for Cygwin that somebody opened
        bugs for in the last week or so.
        * Test MinGW static and shared builds in conjunction with
        APR-util 1.5.x.  (Should be okay; some fixes for shared never
        hit the 1.4.x branch, so it would be good to ensure this is
        finally working.)
        * Check for anything whatsoever that could be merged from
        trunk and see what to do, as presumably there won't be another
        opportunity for a while.

        It would be good for T&R to be no sooner than next Monday.

        Any other thoughts on what to look at before T&R of the first
        1.5.x release?

        --         Born in Roswell... married an alien...
        http://emptyhammock.com/


    I haven't forgotten :)  I want to wrap up the apr_shm stuff on
    Windows and finish working through appropriate fixes that aren't
    in 1.5.x.  Additionally, the traditional Windows build system is
    not ready in the 1.5.x branch but maybe some kind soul will be
    able to take care of that.


BTW, the stuff that seems to be missing from the traditional Windows build system in the 1.5.x branch is

* apr_escape stuff including gen_test_char (I think trunk has this in the Windows build system)

No, I see nothing in trunk for either.

oops, sorry :(
 
I'd like to first rid the 1.5 traditional Windows build of the Release9x & Debug9x targets. Anyone against this?

no concerns here; is there code that can get deleted too?
 

gen_test_char is straight forward enough, generates apr_escape_test_char.h which is included in apr_escape.c.  I am however compelled to think of httpd and this gets tricky when building httpd in it's traditional build (since apr is built during the httpd build) if I need to first build gen_test_char so I can generate the header before compiling the libs. Since this is not needed in APR < 1.5, to do this in httpd may create a requirement for APR 1.5 in httpd. Not sure that's a bad thing.

I'm confused.  httpd still has its own gen_test_char when generates a certain .h, and apr now has one that generates a different .h.  Why wouldn't the build continue to run both programs when building httpd 2.4 in conjunction with apr 1.5?  (I assume that in the future httpd trunk will consume apr_escape and not need both, but we're not there yet.)



There is a existing pre-generated apr_escape_test_char.h in apr/include/private but if differs slightly from the output I get when I run gen_test_char on Windows. What I do get is consistent over the 3 different Windows versions on three different computers and two different versions of MSVC.  The difference between them is;
-    32,30,30,30,30,30,30,30,30,30,31,30,30,30,30,30,30,30,30,30,
-    30,30,30,30,30,30,30,30,30,30,30,30,6,16,63,22,17,22,49,17,
+    32,30,30,30,30,30,30,30,30,30,31,30,30,31,30,30,30,30,30,30,
+    30,30,30,30,30,30,30,30,30,30,30,30,6,16,63,22,17,23,49,17,

What this means is that \r and % are escaped on Windows and OS/2 when performing shell escaping, but are not on other platforms.

Regardless of which I compile with (existing or freshly generated), testescape succeeds. So I seem to have a few options and not sure which way to turn.

1. build and run gen_test_char before compiling APR. This would need to be duplicated in httpd which would create the requirement for APR 1.5.

help me out here (I seem them as completely independent steps which happen to perform a similar function)
 

2. Pre-generate on Windows apr_escape_test_char.h(w) and add to apr/include/private in svn which could be dealt with during the APR build just like apr.h is in Windows (simple and will not affect httpd build).

if committing one for Unix works then that would work too


3. Use the pre-generated one in apr/include/private (simpler and will not affect httpd build).

need those extra chars escaped for shell AFAIK 

I'm leaning toward #2. Comments?

* generated makefiles

Can do.

Regards,
Gregg





--
Born in Roswell... married an alien...
http://emptyhammock.com/