src/ directory
One of my latest problems dealing with APR's multiple directories was find the objects to link. There were two problems with this, one caused by lack of src/, and one partly: 1) If src/ existed in APR, then finding the objects would be a simple "find" call. As it is, I must iterate over all the subdirectories and piecemeal build up the list of objects. [ this is used for linking and date detection ] 2) A simple find does work because of the presence of MM in there. A better organization would be apr/src/ and apr/mm/. That would allow a simple "find" for the src objects, and it would clarify that MM is a subpackage rather than part of the source lump. So... that's my report on dealing with APR's multiple directories, as opposed to APRUTIL's single directory. Of course, my previous comments on simplicity, cleanliness, and understandability of the top-level (when src/ is used) still apply. Cheers, -g -- Greg Stein, http://www.lyra.org/
Re: cvs commit: apr hints.m4 configure.in
David Reid wrote: > > Nope, it *REALLY* didn't work on BeOS the old way. I did some testing and > found that the value wasn't being seen by configure at all and so without > this it's broken. It seems to be some strange thing about the shell on > BeOS, but if you want more info then email me directly... > > With this patch it now builds again and can be seen in action running on > BeOS R5 at > Then I vote we keep it with the patch folded in then. Looks like it works on all required systems. -- === Jim Jagielski [|] [EMAIL PROTECTED] [|] http://www.jaguNET.com/ "Casanova will have many weapons; To beat him you will have to have more than forks and flatulence."
Re: cvs commit: apr hints.m4 configure.in
Nope, it *REALLY* didn't work on BeOS the old way. I did some testing and found that the value wasn't being seen by configure at all and so without this it's broken. It seems to be some strange thing about the shell on BeOS, but if you want more info then email me directly... With this patch it now builds again and can be seen in action running on BeOS R5 at http://bp6.cdws.com.au:8080/ http://bp6.cdws.com.au:8080/cgi-bin/Sensorama david - Original Message - From: "Jim Jagielski" <[EMAIL PROTECTED]> To: <[EMAIL PROTECTED]> Cc: Sent: Saturday, January 13, 2001 1:30 PM Subject: Re: cvs commit: apr hints.m4 configure.in > I'd recommend we back it out then, unless it *really* didn't work for > David the old way... Maybe BeOS and OS/2 weren't including the right > name? > > Brian Havard wrote: > > > > On 12 Jan 2001 17:57:03 -, [EMAIL PROTECTED] wrote: > > > > >dreid 01/01/12 09:57:02 > > > > > > Modified:.hints.m4 configure.in > > > Log: > > > The file_as-socket check wasn't getting picked up correctly, so this > > > corrects that for BeOS and OS/2. Untested for OS/2 so Brian? > > > > Err, well it was working fine for me the way it was. That is, in apr.h I > > get > > > > #define APR_FILES_AS_SOCKETS 0 > > > > It still works after this patch, I just don't know why it was needed. > > > > -- > > __ > > | Brian Havard | "He is not the messiah! | > > | [EMAIL PROTECTED] | He's a very naughty boy!" - Life of Brian | > -- > > > > > -- > === >Jim Jagielski [|] [EMAIL PROTECTED] [|] http://www.jaguNET.com/ > "Casanova will have many weapons; To beat him you will > have to have more than forks and flatulence." >
Re: cvs commit: apr hints.m4 configure.in
I'd recommend we back it out then, unless it *really* didn't work for David the old way... Maybe BeOS and OS/2 weren't including the right name? Brian Havard wrote: > > On 12 Jan 2001 17:57:03 -, [EMAIL PROTECTED] wrote: > > >dreid 01/01/12 09:57:02 > > > > Modified:.hints.m4 configure.in > > Log: > > The file_as-socket check wasn't getting picked up correctly, so this > > corrects that for BeOS and OS/2. Untested for OS/2 so Brian? > > Err, well it was working fine for me the way it was. That is, in apr.h I > get > > #define APR_FILES_AS_SOCKETS 0 > > It still works after this patch, I just don't know why it was needed. > > -- > > __ > | Brian Havard | "He is not the messiah! > | > | [EMAIL PROTECTED] | He's a very naughty boy!" - Life of Brian | > > -- > -- === Jim Jagielski [|] [EMAIL PROTECTED] [|] http://www.jaguNET.com/ "Casanova will have many weapons; To beat him you will have to have more than forks and flatulence."
Re: cvs commit: apr-util STATUS
> wrowe 01/01/12 20:23:31 > > Modified:.STATUS > Log: > Vote on some things so we get this out the door! > >RELEASE SHOWSTOPPERS: > > +* This vote ends on Wednesday! > + Source code layout needs to be decided for apr and apr-util. There > + has been some discussion about adding a src/ directory to apr or > + removing it from apr-util/. Either way, this needs to be decided. > + Those for adding apr/src/: gstein > + Those removing apr-util/src/ : wrowe, rbb Sorry, let's do this in one place, apr/STATUS ... we don't have two projects and mail lists going on. >* Prefix for the code. > Status: >APR: Ryan +1, Greg +1 > +Will asks "Is that apu_ prefix? I'm +++1 for that! Same timetable to deal with this. > >* Compatibility header to deal with renamed symbols. The only 'right' way is to offer an ap_compat.h apache symbols, which includes: apu_compat.h apr-util symbols, which includes: apr_compat.h Very simple, I'll implement (if it wasn't already done) Sunday night.
Re: cvs commit: apr hints.m4 configure.in
On 12 Jan 2001 17:57:03 -, [EMAIL PROTECTED] wrote: >dreid 01/01/12 09:57:02 > > Modified:.hints.m4 configure.in > Log: > The file_as-socket check wasn't getting picked up correctly, so this > corrects that for BeOS and OS/2. Untested for OS/2 so Brian? Err, well it was working fine for me the way it was. That is, in apr.h I get #define APR_FILES_AS_SOCKETS 0 It still works after this patch, I just don't know why it was needed. -- __ | Brian Havard | "He is not the messiah! | | [EMAIL PROTECTED] | He's a very naughty boy!" - Life of Brian | --
Re: FreeBSD version check for thread-safe sendfile
At 11 Jan 2001 16:12:02 -0500, Jeff Trawick wrote: > If there is 3.4.x, then perhaps it should be 340 for 3.4, 341 for > 3.4.1, etc. > > The version could be set to > > uname -r | sed -e 's/\(.\)\.\(.\)\.\(.\).*/\1\2\3/' | sed -e > 's/\(.\)\.\(.\)\-.*/\1\20/' > > yielding 341 for 3.4.1-RELEASE and 340 for 3.4-RELEASE. > > Why limit the variable to FreeBSD? We'll have (or already have) > similar checks on other platforms. Make the variable name more > generic. If version specific hack for FreeBSD is required, is it useful to use "sysctl kern.osreldate"? % sysctl kern.osreldate kern.osreldate: 500015 # See http://www.FreeBSD.org/porters-handbook/porting-versions.html # for detail of values -- Jun Kuriyama <[EMAIL PROTECTED]> // IMG SRC, Inc. <[EMAIL PROTECTED]> // FreeBSD Project