Conditional in Makefiles

Hi Niels, hi Sjoerd,

Based on a condition in configure I want to compile different files.

I know that it is possible to include a directory based on a condition, 
but couldn't find a solution for files. Is there any?

-- If that's not possible is there some magic for the Makefile.ag files 
to write a snippet of code that survives autoconf and automake without 
changes?

Jan

--

-- 
Jan Rittinger
Database Systems
Technische Universität München (Germany)
http://www-db.in.tum.de/~rittinge/

-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV

Re: Conditional in Makefiles

On 2007-01-15 15:54, Jan Rittinger wrote:
> Hi Niels, hi Sjoerd,
> 
> Based on a condition in configure I want to compile different files.
> 
> I know that it is possible to include a directory based on a condition, 
> but couldn't find a solution for files. Is there any?
> 
> -- If that's not possible is there some magic for the Makefile.ag files 
> to write a snippet of code that survives autoconf and automake without 
> changes?
> 
> Jan
> 

Look for COND in Makefile.ag.  There are some that have that, and that's
what you can use.  There's even one in pathfinder in src/tools/Makefile.ag.

--

-- 
Sjoerd Mullender

-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV

_______________________________________________

Re: [Monetdb-pf-checkins] pathfinder/compiler/include abssynprint.h, 1.9.4.1, 1.9.4.2 coreprint.h, 1.8.4.1, 1.8.4.2

Hi Peter,

I hope the functions PFaprint and PFpcore are for your personal 
debugging only. If that's the case please remove them. (Otherwise I do 
not understand why you introduced these functions.)

Jan

On 01/13/2007 03:40 AM, Peter Boncz wrote with possible deletions:
> Update of /cvsroot/monetdb/pathfinder/compiler/include
> In directory sc8-pr-cvs7.sourceforge.net:/tmp/cvs-serv27995/compiler/include
> 
> Modified Files:
>       Tag: XQuery_0-14
> 	abssynprint.h coreprint.h 
> 
> Index: coreprint.h
> ===================================================================
> RCS file: /cvsroot/monetdb/pathfinder/compiler/include/coreprint.h,v
> retrieving revision 1.8.4.1
> retrieving revision 1.8.4.2
> diff -u -d -r1.8.4.1 -r1.8.4.2
> --- coreprint.h	3 Jan 2007 13:21:18 -0000	1.8.4.1
> +++ coreprint.h	13 Jan 2007 02:40:47 -0000	1.8.4.2
>  <at>  <at>  -48,6 +48,8  <at>  <at> 
>  
>  void PFcore_pretty (FILE *f, PFcnode_t *root);
>  
> +void PFpcore (PFcnode_t *root);
> +

Re: Conditional in Makefiles

Thanks Sjoerd for your answer. I think COND is the one I was looking 
for. However defining a library twice with different conditions ignores 
the second condition. Is their also the possibility for EITHER/OR (in 
the following example HAVE_BAR?bar.c:bar_alternative.c)?

I tried it the following way (where HAVE_BAR is not HAVE_NO_BAR):
lib_foo = {
         COND = HAVE_BAR
         NOINST
         DIR = libdir
         SOURCES = \
                 foo1.c \
                 foo2.c \
                 bar.c
}

lib_foo = {
         COND = HAVE_NO_BAR
         NOINST
         DIR = libdir
         SOURCES = \
                 foo1.c \
                 foo2.c \
                 bar_alternative.c
}

On 01/15/2007 04:16 PM, Sjoerd Mullender wrote with possible deletions:
> On 2007-01-15 15:54, Jan Rittinger wrote:
>> Hi Niels, hi Sjoerd,
>>

Re: Conditional in Makefiles

On 2007-01-15 18:08, Jan Rittinger wrote:
> Thanks Sjoerd for your answer. I think COND is the one I was looking 
> for. However defining a library twice with different conditions ignores 
> the second condition. Is their also the possibility for EITHER/OR (in 
> the following example HAVE_BAR?bar.c:bar_alternative.c)?
> 
> I tried it the following way (where HAVE_BAR is not HAVE_NO_BAR):
> lib_foo = {
>          COND = HAVE_BAR
>          NOINST
>          DIR = libdir
>          SOURCES = \
>                  foo1.c \
>                  foo2.c \
>                  bar.c
> }
> 
> lib_foo = {
>          COND = HAVE_NO_BAR
>          NOINST
>          DIR = libdir
>          SOURCES = \
>                  foo1.c \
>                  foo2.c \
>                  bar_alternative.c
> }

I don't think we have support for anything like this.  The only thing I
can think of is a C file which chooses which of the implementations you
want with an #ifdef (and if you want to keep the implementations in

MonetDB is at SourceForge

The transition of monet5 to SourceForge is complete.  To get it do one
of (each command is one line):

cvs -d :ext:<name> <at> monetdb.cvs.sourceforge.net:/cvsroot/monetdb co MonetDB5

cvs -d :pserver:anonymous@...:/cvsroot/monetdb
co MonetDB5

The former requires that you have developer access to the MonetDB
project at SourceForge and you need to fill in your SourceForge account.
 The latter is an anonymous, read-only checkout.

If you have changes in your old CWI-based copy, do the following:
cd .../monet5 # (go to your old copy from CWI)
cvs update
cvs diff -c > $HOME/m5-diffs

cd .../MonetDB5 # (go to the freshly checked out copy from SF)
patch -p0 < $HOME/m5-diffs

After this you can look at what you changed, correct any conflicts, and
commit.

Enjoy!

--

-- 
Sjoerd Mullender

-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT

Re: [Monetdb-checkins] MonetDB5/src/mal mal_client.mx, 1.139, 1.140 mal_session.mx, 1.129, 1.130

On Wed, Jan 17, 2007 at 07:13:52AM +0000, Niels Nes wrote:
> Update of /cvsroot/monetdb/MonetDB5/src/mal
> In directory sc8-pr-cvs7.sourceforge.net:/tmp/cvs-serv9203/src/mal
> 
> Modified Files:
> 	mal_client.mx mal_session.mx 
> Log Message:
> increased number of client slots from 32 to 64
> don't crash on running out ot client slots 

for info: M4 uses/allows even 512 clients:

./MonetDB4/src/monet/monet_client.mx:#define MAXCLIENTS 512

Stefan

--

-- 
| Dr. Stefan Manegold | mailto:Stefan.Manegold@... |
| CWI,  P.O.Box 94079 | http://www.cwi.nl/~manegold/  |
| 1090 GB Amsterdam   | Tel.: +31 (20) 592-4212       |
| The Netherlands     | Fax : +31 (20) 592-4312       |

-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV

Re: [Monetdb-pf-checkins] pathfinder/compiler/algebra/map intro_proxy.c, 1.19, 1.20

On Tue, Jan 16, 2007 at 01:23:11PM +0000, Jan Rittinger wrote:
> Update of /cvsroot/monetdb/pathfinder/compiler/algebra/map
> In directory sc8-pr-cvs7.sourceforge.net:/tmp/cvs-serv16649/algebra/map
> 
> Modified Files:
> 	intro_proxy.c 
> Log Message:
> -- Perhaps this change keeps icc quiet...

It does.
Thank you very much!

Stefan

> Index: intro_proxy.c
> ===================================================================
> RCS file: /cvsroot/monetdb/pathfinder/compiler/algebra/map/intro_proxy.c,v
> retrieving revision 1.19
> retrieving revision 1.20
> diff -u -d -r1.19 -r1.20
> --- intro_proxy.c	4 Jan 2007 15:26:18 -0000	1.19
> +++ intro_proxy.c	16 Jan 2007 13:23:07 -0000	1.20
>  <at>  <at>  -2193,7 +2193,7  <at>  <at> 
>          switch (p->kind) {
>              case la_attach:
>                  collect_mappings_worker (p->sem.attach.attname,
> -                                         PFalg_attlist (),
> +                                         PFalg_attlist_ (0, NULL),
>                                           req_count,
>                                           req_col_names,

incorrectly marked keyed

I don't really understand the monetdb warning below, which I get when
running test cases for pf/tijah. it's actually nice that monetdb
performs all these BATpropchecks when running the test cases, but what's
actually the problem here? I printed the BAT and its info() now to check
myself, but can't find any problem. A void-headed BAT will always be
keyed, or not?

best -henning

# Shredded 1 XML document (dilbert.xml), total time after commit=0.007s
!WARNING: BATpropcheck: BAT tj_testcoll1_size1(-79) was incorrectly
marked keyed!
bat("tj_testcoll1_size1").info().print();
#-------------------------------------------------#
# h                     t                         # name
# str                   str                       # type
#-------------------------------------------------#
[ "version",              "25105"                 ]
[ "batId",                "tj_testcoll1_size1"    ]
[ "batCacheid",           "79"                    ]
[ "batParentid",          "0"                     ]
[ "batSharecnt",          "0"                     ]
[ "head",                 "void"                  ]
[ "tail",                 "int"                   ]
[ "batPersistence",       "persistent"            ]
[ "batRestricted",        "read-only"             ]
[ "batRefcnt",            "1"                     ]
[ "batLRefcnt",           "2"                     ]
[ "batDirty",             "clean"                 ]
[ "batSet",               "0"                     ]

Re: incorrectly marked keyed

Henning,

- you print the BAT *after* BATpropcheck has complained about (and fixed!)
  wrong properties

- it might be either head or tail that BATpropcheck complains about 
  (internally, it checks only on head, but it's called for B & reverse(B))

- void columns are only key in case the seqbase is not NIL

- who/which code generates the BAT?

Stefan

On Wed, Jan 17, 2007 at 09:04:55AM +0100, Henning Rode wrote:
> I don't really understand the monetdb warning below, which I get when
> running test cases for pf/tijah. it's actually nice that monetdb
> performs all these BATpropchecks when running the test cases, but what's
> actually the problem here? I printed the BAT and its info() now to check
> myself, but can't find any problem. A void-headed BAT will always be
> keyed, or not?
> 
> best -henning
> 
> # Shredded 1 XML document (dilbert.xml), total time after commit=0.007s
> !WARNING: BATpropcheck: BAT tj_testcoll1_size1(-79) was incorrectly
> marked keyed!
> bat("tj_testcoll1_size1").info().print();
> #-------------------------------------------------#
> # h                     t                         # name