plplot-cvs

[Plplot-cvs] CVS update notice (plplot/cf)

[Maurice L. <ml...@us...>]

Update of /cvsroot/plplot/plplot/cf
In directory usw-pr-cvs1:/tmp/cvs-serv15652

Modified Files:
	sysconf.in 
Log Message:
Don't enable the --thread_safe flag automatically for KCC any more.
Better to just let the user set it in a site config file if desired.

[Plplot-cvs] CVS update notice (plplot/cf)

[Geoffrey F. <fu...@us...>]

Update of /cvsroot/plplot/plplot/cf
In directory usw-pr-cvs1:/tmp/cvs-serv4666/cf

Modified Files:
	configure.in dist.in initlib.in inst_lib.in install.in 
	lib_sh_linux.in misc.in plConfig.h.in plDevs.h.in 
Added Files:
	.cvsignore dyndrv.in 
Log Message:
Merged dyndrv_1 branch to HEAD.  More detailed descriptions of all the
mini steps were contained in the commit messages along that branch.
A brief summary of the major points:

1) configure has been modified to synthesize the set of driver modules
to compile.  Each driver is compiled either as a static module, or a
dynamic module.  Separate lists hold drivers in each category.  Only
the selected driver modules are compiled.  Disabled driver modules are
not compiled.  

2) Each driver may provide one or more devices.

3) The dispatch table is dynamically initialized, first with the list
of static drivers (and devices), and then with the set of dynamic
drivers.  Then this list is sorted, based on a presentation order
field in the dispatch entry data structure.

4) PLStream now has a member for the associated device.  The global
dispatch table is abolsihed, and instead the core routines vector to
the device routines by bouncing throught the dispatch table associated
with the stream.

5) Each device has a new required function which is responsible for
initializing the per-device dispatch table with the proper function
addresses.  Only this dispatch_init function is looked up using the
dynamic symbol search stuff.  This function must have a stylized name
so that it can be programmatically located, but after that, the names
of the device-specific entry points are arbitrary (provided they have
the proper signature).

[Plplot-cvs] CVS update notice (plplot/cf)

[Geoffrey F. <fu...@us...>]

Update of /cvsroot/plplot/plplot/cf
In directory usw-pr-cvs1:/tmp/cvs-serv1026/cf

Modified Files:
	configure.in initlib.in install.in misc.in sysloc.in 
Added Files:
	pkg_java.in 
Log Message:
First attempt to enable support for a Java binding for PLplot.
Activate via --enable-java, and make sure you have JAVA_HOME set to
the root of your JDK.

This much work on the configuration system is enough to get as far as
building the embrionic Java support in a Javoid (Javonian?) directory
structure.  IIRC, Java requires files to be laid out on disk in the
structure which the package import system will see.  Mapping this onto
the PLplot language porting conventions in CVS, is done, as usual,
with soft links.  

This should be regarded as an initial point of discussion, as the
construction of a top-rate Java binding looks to be a substantial
undertaking.

[Plplot-cvs] CVS update notice (plplot/cf)

[Geoffrey F. <fu...@us...>]

Update of /cvsroot/plplot/plplot/cf
In directory usw-pr-cvs1:/tmp/cvs-serv24158/cf

Modified Files:
	pkg_java.in 
Log Message:
Rework the java compilation pattern rule framework so that java
components are compiled from the tmp dir, with the resulting class
files stored under the tmp/java tree.  This is because we need the
file name that is compiled by make to be the same as the Emacs buffer
name, in order for auto-error placement to work.  The previous scheme
worked too, but Emacs couldn't tell that the file being compiled was
the same as the file currently mapped to a buffer, so you always
winded up editing different copies of the same file.  Blecherous.

This new compilation strategy may allow further simplification of the
way the softlink configuration is done, but that's defered for now.

[Plplot-cvs] CVS update notice (plplot/cf)

[Geoffrey F. <fu...@us...>]

Update of /cvsroot/plplot/plplot/cf
In directory usw-pr-cvs1:/tmp/cvs-serv5657/cf

Modified Files:
	dyndrv.in 
Log Message:
Fix very tricky error in building of the dynloadable drivers.

C programs like x01c were working fine with dynamic drivers, but only
accidentally.  Specifically, references by the drivers to functions
provided by libplplot, were being resolved in some cases as a result
of the peculiarities of how the final executable was being built.

However, in the case of the java binding, we don't control the way in
which Java's System.loadLibrary( "plplot" ); manages the interraction
with dlopen(), so we have to prime the pump.  To make it all work, we
have to actually link the drivers against libplplot.so, so that when
the drivers are dynloaded into libplplot during the Java static init
section in PLStream.java, they will actually resolve symbols against
the (already loaded) libplplot.so.  

It all comes down to understanding dlopen() and friends.  Evidently
when the JVM does the external (native) load, it won't use
RTLD_GLOBAL.  I guess that makes some kind of sense from a security
standpoint, although it complicates the situation for dynloadable
libraries which want to dynload additional components.  

Anyway, it works now.  The dynamic drivers can be loaded into
executable programs compiled in the usual way (shown by the PLplot
makefile), or into libplplot.so when libplplot.so is itself dynloaded
into some other process, such as occurs in the case of the JVM loading
the PLplot JNI binding dynamically.

Cool.

[Plplot-cvs] CVS update notice (plplot/cf)

[Geoffrey F. <fu...@us...>]

Update of /cvsroot/plplot/plplot/cf
In directory usw-pr-cvs1:/tmp/cvs-serv15186/cf

Modified Files:
	dyndrv.in 
Log Message:
Revise text message at the end of the dynamic drivers target.

[Plplot-cvs] CVS update notice (plplot/cf)

[Geoffrey F. <fu...@us...>]

Update of /cvsroot/plplot/plplot/cf
In directory usw-pr-cvs1:/tmp/cvs-serv15655/cf

Modified Files:
	pkg_java.in 
Log Message:
Minor correction to specification order of Java components to avoid
spurious build failure.  This should be done more properly eventually,
but this is better than how it was.

Also correct final text output message.

[Plplot-cvs] CVS update notice (plplot/cf)

[Geoffrey F. <fu...@us...>]

Update of /cvsroot/plplot/plplot/cf
In directory usw-pr-cvs1:/tmp/cvs-serv7566/cf

Modified Files:
	inst_lib.in 
Log Message:
s/dispatch.h/disptab.h/

[Plplot-cvs] CVS update notice (plplot/cf)

[Geoffrey F. <fu...@us...>]

Update of /cvsroot/plplot/plplot/cf
In directory usw-pr-cvs1:/tmp/cvs-serv31051/cf

Modified Files:
	pkg_java.in 
Log Message:
Reworked the java targets a bit so that we don't compile the java
examples when we do a default build.  Ask for jdemos, or jall, as
appropriate.

[Plplot-cvs] CVS update notice (plplot/cf)

[Geoffrey F. <fu...@us...>]

Update of /cvsroot/plplot/plplot/cf
In directory usw-pr-cvs1:/tmp/cvs-serv27473/cf

Modified Files:
	dyndrv.in 
Log Message:
Add a blank line at the end, so that when configure composes Makefile
by catting everything together, the top of the next file doesn't show
up as part of the last rule in this file.

[Plplot-cvs] CVS update notice (plplot/cf)

[Geoffrey F. <fu...@us...>]

Update of /cvsroot/plplot/plplot/cf
In directory usw-pr-cvs1:/tmp/cvs-serv28350/cf

Modified Files:
	lib_sh_linux.in 
Log Message:
Remarkably, somehow, the shared aux libs weren't dependent upon the
.o's in the shared subdir, so make was not compiling the -fPIC
versions.  I don't see how this could've been missed.  There must've
been additional, unobvious factors impinging on this situation.

[Plplot-cvs] CVS update notice (plplot/cf)

[Geoffrey F. <fu...@us...>]

Update of /cvsroot/plplot/plplot/cf
In directory usw-pr-cvs1:/tmp/cvs-serv29824/cf

Modified Files:
	pkg_tcl.in 
Log Message:
MAT_OBJS_SO wasn't set.  I don't see how it was working before.  There
must've been a compensating mistake somewhere that I must've
obliterated in the dyndrv merge.  Hmm.  Anyway, it works (builds
libtclmatrix) now.

[Plplot-cvs] CVS update notice (plplot/cf)

[Geoffrey F. <fu...@us...>]

Update of /cvsroot/plplot/plplot/cf
In directory usw-pr-cvs1:/tmp/cvs-serv31239/cf

Modified Files:
	pkg_tk.in 
Log Message:
Remove tk.c from the set of sources for the Tk extension support.  It
is now listed in the static drivers list, only if configured with Tk
support.  Before, tk.o was being supplied twice during lib builds, if
Tk was enabled, which resulted in multiply defined symbol errors.

[Plplot-cvs] CVS update notice (plplot/cf)

[Geoffrey F. <fu...@us...>]

Update of /cvsroot/plplot/plplot/cf
In directory usw-pr-cvs1:/tmp/cvs-serv26674/cf

Modified Files:
	dyndrv.in 
Log Message:
Add $(MAT_LINK) $(LDC_LIBS) when linking dynamic drivers.  This is the
same basic issue as when we previously discovered we had to link
against libplplot, in case the host app didn't use the most lenient
combination of RTLD_* flags in its dlopen call to load libplplot.  The
point is, the dynamic drivers not only depend on libplplot, but
thereby vicariously also depend on the other libs that libplplot
depends on, so all of this has to be supplied to the linker when
forming the dynloadable driver images.

Maybe this will get simpler as we progress.  Perhaps once all drivers
are dynamic, then maybe we'll reach the point where libplplot won't
actually depend on any additional libraries, and only the individual
(dynamic) drivers will have linkage requirements to external libs.
But until we can get that advanced with the makefiles, we'll just have
to supply all these lib specs when forming the dynloadable driver
images. 

[Plplot-cvs] CVS update notice (plplot/cf)

[Geoffrey F. <fu...@us...>]

Update of /cvsroot/plplot/plplot/cf
In directory usw-pr-cvs1:/tmp/cvs-serv26920/cf

Modified Files:
	pkg_java.in 
Log Message:
First (weak) attempt at an install target for the Java binding and
examples.

[Plplot-cvs] CVS update notice (plplot/cf)

[Joao C. <jc...@us...>]

Update of /cvsroot/plplot/plplot/cf
In directory usw-pr-cvs1:/tmp/cvs-serv13687/cf

Modified Files:
	initlib.in 
Log Message:
GNOME_OBJ are now in static_objects; cf/pkg_gnome.in can dissapear?

[Plplot-cvs] CVS update notice (plplot/cf)

[Joao C. <jc...@us...>]

Update of /cvsroot/plplot/plplot/cf
In directory usw-pr-cvs1:/tmp/cvs-serv13969/cf

Modified Files:
	configure.in dyndrv.in 
Log Message:
Add dynamic loading support for all but the xwin, tk, linuxvga and NeXt drivers.

[Plplot-cvs] CVS update notice (plplot/cf)

[Alan W. I. <ai...@us...>]

Update of /cvsroot/plplot/plplot/cf
In directory usw-pr-cvs1:/tmp/cvs-serv11339

Modified Files:
	dyndrv.in 
Log Message:
Fixed library name problem by replacing explicit reference with symbolic
reference, i.e., -lplplot by -l$(PLLIB_NAME)$(LIB_TAG)

[Plplot-cvs] CVS update notice (plplot/cf)

[Alan W. I. <ai...@us...>]

Update of /cvsroot/plplot/plplot/cf
In directory usw-pr-cvs1:/tmp/cvs-serv27821

Modified Files:
	sysconf.in 
Log Message:
For Linux always use -ansi -pedantic -Wall for compilation of C code.
(implemented with SYS_FLAGS_C="-ansi -pedantic -Wall")

[Plplot-cvs] CVS update notice (plplot/cf)

[Alan W. I. <ai...@us...>]

Update of /cvsroot/plplot/plplot/cf
In directory usw-pr-cvs1:/tmp/cvs-serv14148

Modified Files:
	sysconf.in 
Log Message:
Change from -ansi -pedantic -Wall to -ansi for now

[Plplot-cvs] CVS update notice (plplot/cf)

[Alan W. I. <ai...@us...>]

Update of /cvsroot/plplot/plplot/cf
In directory usw-pr-cvs1:/tmp/cvs-serv2810

Modified Files:
	configure.in 
Log Message:
Fixed bug in specification of dynamic versions of png and jpeg

[Plplot-cvs] CVS update notice (plplot/cf)

[Joao C. <jc...@us...>]

Update of /cvsroot/plplot/plplot/cf
In directory usw-pr-cvs1:/tmp/cvs-serv18553/cf

Modified Files:
	configure.in 
Log Message:
Give the user some feedback during driver/device detection.
Change the way drivers.db is generated.
Use DYNAMIC_DRIVERS instead of PL_DYNAMIC_DRIVER_LIST during drivers.db generation.
Alan, I have undone your last change; together with a change in gd.c I think the png/jpeg will work again.

[Plplot-cvs] CVS update notice (plplot/cf)

[Joao C. <jc...@us...>]

Update of /cvsroot/plplot/plplot/cf
In directory usw-pr-cvs1:/tmp/cvs-serv30565/cf

Modified Files:
	configure.in 
Log Message:
oops, in the last commit I forgot to fix the gd entry when creating drivers.db

[Plplot-cvs] CVS update notice (plplot/cf)

[Joao C. <jc...@us...>]

Update of /cvsroot/plplot/plplot/cf
In directory usw-pr-cvs1:/tmp/cvs-serv9174/cf

Modified Files:
	configure.in 
Log Message:
Allow per-device selection in the gd dyn-driver case when generating driver.db

[Plplot-cvs] CVS update notice (plplot/cf)

[Joao C. <jc...@us...>]

Update of /cvsroot/plplot/plplot/cf
In directory usw-pr-cvs1:/tmp/cvs-serv28684/cf

Modified Files:
	plDevs.h.in 
Log Message:
Add support for the "ntk" driver.