libphidget-devel — Mailing list for developers of libphidget

Re: [Libphidget-devel] just playing with indent . . .

[Vadim T. <vt...@fr...>]

According to Jack Strohm:

> > Here's an update: I'm looking at the Apache source code - those guys made it
> > work. The key seems to be -T option, I'll play with that and let's see what
> > happens.
> 
> ok.
> 
> > 
> > Some more: if you could read 'man indent' and experiment with the detailed
> > options a little bit, I'd feel better about the style. Granted, neither of
> > the default styles (K&R, GNU, BSD) is not nice enough as of today.
> 
> I'll see what I come up with

Meanwhile, I'll do this: branch this stuff into INDENT branch, and keep
playing there, so you don't have to suffer and can just keep working on the
code. When the indent stuff is done and over with, the branch will either
get discarded if it ends up in a total failure, or merged back (actually,
just the makefiles and indent options have to be merged back).

I should've done this in the first place...

--vt

Re: [Libphidget-devel] just playing with indent . . .

[Jack S. <js...@ja...>]

On Mon, 2002-09-09 at 21:42, Vadim Tkachenko wrote:
> According to Jack Strohm:
> 
> > > > hmmm, doesn't sound good.
> > > 
> > > OK, preliminary research shows that I've made a major flop here :( Shouldn't
> > > have touched the indent at all. So far, I haven't found the tools that work
> > > reliably (astyle has its own problems, and there's nothing else, really...)
> > > 
> > > Two ways out:
> > > 
> > > 1. Roll back all the cc/c/h commits to the way it was before
> > > 
> > > 2. Curse a lot, and keep the code the way it is now, and remove (rather
> > >    comment out, though) the indent related portions of the makefile, hoping
> > >    silently that there'll be a day when they fix their bugs.
> > > 
> > > Even though I hate the GNU style, I'd rather go with #2, 'cause this is at
> > > least a sort of a standard... Minor mistakes that indent introduced can be
> > > fixed manually, should there be a need to do so.
> > > 
> > > #2 gets my vote. What's yours? Remember, you're the boss here.
> > 
> > I'd say #2, although, the current way the code is indented needs a bit
> > of work.  What is GNU style exactly anyway.
> 
> GNU style is what you see ;) The current contents of ./indent.rules is
> simply '--gnu-style', therefore, piping the source through indent produces
> the code formatted according to GNU style.

well, I don't like GNU, it messes up the comments a lot.

> 
> Here's an update: I'm looking at the Apache source code - those guys made it
> work. The key seems to be -T option, I'll play with that and let's see what
> happens.

ok.

> 
> Some more: if you could read 'man indent' and experiment with the detailed
> options a little bit, I'd feel better about the style. Granted, neither of
> the default styles (K&R, GNU, BSD) is not nice enough as of today.

I'll see what I come up with

Re: [Libphidget-devel] just playing with indent . . .

[Vadim T. <vt...@fr...>]

According to Jack Strohm:

> > > hmmm, doesn't sound good.
> > 
> > OK, preliminary research shows that I've made a major flop here :( Shouldn't
> > have touched the indent at all. So far, I haven't found the tools that work
> > reliably (astyle has its own problems, and there's nothing else, really...)
> > 
> > Two ways out:
> > 
> > 1. Roll back all the cc/c/h commits to the way it was before
> > 
> > 2. Curse a lot, and keep the code the way it is now, and remove (rather
> >    comment out, though) the indent related portions of the makefile, hoping
> >    silently that there'll be a day when they fix their bugs.
> > 
> > Even though I hate the GNU style, I'd rather go with #2, 'cause this is at
> > least a sort of a standard... Minor mistakes that indent introduced can be
> > fixed manually, should there be a need to do so.
> > 
> > #2 gets my vote. What's yours? Remember, you're the boss here.
> 
> I'd say #2, although, the current way the code is indented needs a bit
> of work.  What is GNU style exactly anyway.

GNU style is what you see ;) The current contents of ./indent.rules is
simply '--gnu-style', therefore, piping the source through indent produces
the code formatted according to GNU style.

Here's an update: I'm looking at the Apache source code - those guys made it
work. The key seems to be -T option, I'll play with that and let's see what
happens.

Some more: if you could read 'man indent' and experiment with the detailed
options a little bit, I'd feel better about the style. Granted, neither of
the default styles (K&R, GNU, BSD) is not nice enough as of today.

--vt

Re: [Libphidget-devel] just playing with indent . . .

[Jack S. <js...@ja...>]

On Mon, 2002-09-09 at 20:50, Vadim Tkachenko wrote:
> According to Jack Strohm:
> 
> > hmmm, doesn't sound good.
> 
> OK, preliminary research shows that I've made a major flop here :( Shouldn't
> have touched the indent at all. So far, I haven't found the tools that work
> reliably (astyle has its own problems, and there's nothing else, really...)
> 
> Two ways out:
> 
> 1. Roll back all the cc/c/h commits to the way it was before
> 
> 2. Curse a lot, and keep the code the way it is now, and remove (rather
>    comment out, though) the indent related portions of the makefile, hoping
>    silently that there'll be a day when they fix their bugs.
> 
> Even though I hate the GNU style, I'd rather go with #2, 'cause this is at
> least a sort of a standard... Minor mistakes that indent introduced can be
> fixed manually, should there be a need to do so.
> 
> #2 gets my vote. What's yours? Remember, you're the boss here.

I'd say #2, although, the current way the code is indented needs a bit
of work.  What is GNU style exactly anyway.

Re: [Libphidget-devel] just playing with indent . . .

[Vadim T. <vt...@fr...>]

According to Jack Strohm:

> hmmm, doesn't sound good.

OK, preliminary research shows that I've made a major flop here :( Shouldn't
have touched the indent at all. So far, I haven't found the tools that work
reliably (astyle has its own problems, and there's nothing else, really...)

Two ways out:

1. Roll back all the cc/c/h commits to the way it was before

2. Curse a lot, and keep the code the way it is now, and remove (rather
   comment out, though) the indent related portions of the makefile, hoping
   silently that there'll be a day when they fix their bugs.

Even though I hate the GNU style, I'd rather go with #2, 'cause this is at
least a sort of a standard... Minor mistakes that indent introduced can be
fixed manually, should there be a need to do so.

#2 gets my vote. What's yours? Remember, you're the boss here.

--vt

Re: [Libphidget-devel] just playing with indent . . .

[Vadim T. <vt...@fr...>]

According to Jack Strohm:

> > > Maybe since we are running different versions, the default settings are
> > > different.  Maybe we should explicitly set all options for the default.
> > 
> > Nope,
> > 
> > Here's the root cause:
> > 
> > http://home.hccnet.nl/d.ingamells/indent_15.html#SEC15
> > 
> > It's just not smart enough. Funny, I've been using it with Java files quite
> > consistently... Probably because my rules were simpler.
> > 
> > Give me some time, I'll sort it out.
> > 
> 
> hmmm, doesn't sound good.

Well, the good news is that whenever indent starts working right, we're
ready ;)

Meanwhile, I'm looking at astyle and beyond... Sheesh.

--vt

Re: [Libphidget-devel] just playing with indent . . .

[Jack S. <js...@ja...>]

On Mon, 2002-09-09 at 19:26, Vadim Tkachenko wrote:
> According to Jack Strohm:
> > On Mon, 2002-09-09 at 19:19, Vadim Tkachenko wrote:
> > > According to Jack Strohm:
> > > 
> > > > which version of indent are you using?
> > >  
> > > > [jstrohm@oishi libphidget]$ indent --version
> > > > GNU indent 2.2.8
> > > 
> > > 2.2.5 here.
> > > 
> > > This is nuts... Do this: just remove all .c, .h and .cc files and do 'cvs
> > > update' again (hope you didn't change anything), and just hack away on that.
> > > Meanwhile, I'll investigate the alternatives to indent, and/or find a
> > > combination of options that produces an idempotent processing.
> > 
> > that's what I've been doing, "cvs release", and then a full checkout.
> > 
> > Maybe since we are running different versions, the default settings are
> > different.  Maybe we should explicitly set all options for the default.
> 
> Nope,
> 
> Here's the root cause:
> 
> http://home.hccnet.nl/d.ingamells/indent_15.html#SEC15
> 
> It's just not smart enough. Funny, I've been using it with Java files quite
> consistently... Probably because my rules were simpler.
> 
> Give me some time, I'll sort it out.
> 

hmmm, doesn't sound good.

Re: [Libphidget-devel] just playing with indent . . .

[Vadim T. <vt...@fr...>]

According to Jack Strohm:
> On Mon, 2002-09-09 at 19:19, Vadim Tkachenko wrote:
> > According to Jack Strohm:
> > 
> > > which version of indent are you using?
> >  
> > > [jstrohm@oishi libphidget]$ indent --version
> > > GNU indent 2.2.8
> > 
> > 2.2.5 here.
> > 
> > This is nuts... Do this: just remove all .c, .h and .cc files and do 'cvs
> > update' again (hope you didn't change anything), and just hack away on that.
> > Meanwhile, I'll investigate the alternatives to indent, and/or find a
> > combination of options that produces an idempotent processing.
> 
> that's what I've been doing, "cvs release", and then a full checkout.
> 
> Maybe since we are running different versions, the default settings are
> different.  Maybe we should explicitly set all options for the default.

Nope,

Here's the root cause:

http://home.hccnet.nl/d.ingamells/indent_15.html#SEC15

It's just not smart enough. Funny, I've been using it with Java files quite
consistently... Probably because my rules were simpler.

Give me some time, I'll sort it out.

--vt

Re: [Libphidget-devel] just playing with indent . . .

[Jack S. <js...@ja...>]

On Mon, 2002-09-09 at 19:19, Vadim Tkachenko wrote:
> According to Jack Strohm:
> 
> > which version of indent are you using?
>  
> > [jstrohm@oishi libphidget]$ indent --version
> > GNU indent 2.2.8
> 
> 2.2.5 here.
> 
> This is nuts... Do this: just remove all .c, .h and .cc files and do 'cvs
> update' again (hope you didn't change anything), and just hack away on that.
> Meanwhile, I'll investigate the alternatives to indent, and/or find a
> combination of options that produces an idempotent processing.

that's what I've been doing, "cvs release", and then a full checkout.

Maybe since we are running different versions, the default settings are
different.  Maybe we should explicitly set all options for the default.

Re: [Libphidget-devel] just playing with indent . . .

[Vadim T. <vt...@fr...>]

According to Jack Strohm:

> which version of indent are you using?
 
> [jstrohm@oishi libphidget]$ indent --version
> GNU indent 2.2.8

2.2.5 here.

This is nuts... Do this: just remove all .c, .h and .cc files and do 'cvs
update' again (hope you didn't change anything), and just hack away on that.
Meanwhile, I'll investigate the alternatives to indent, and/or find a
combination of options that produces an idempotent processing.

--vt

Re: [Libphidget-devel] just playing with indent . . .

[Jack S. <js...@ja...>]

which version of indent are you using?

[jstrohm@oishi libphidget]$ indent --version
GNU indent 2.2.8

Re: [Libphidget-devel] just playing with indent . . .

[Jack S. <js...@ja...>]

On Mon, 2002-09-09 at 19:02, Vadim Tkachenko wrote:
> According to Jack Strohm:
> 
> > Now with indent support I did this:
> > 
> > cvs update
> > make indent-local
> > make indent
> > cvs commit
> > 
> > And I get this:
> > 
> > CVS:
> > ----------------------------------------------------------------------
> > CVS: Enter Log.  Lines beginning with `CVS:' are removed automatically
> > CVS:
> > CVS: Committing in .
> > CVS:
> > CVS: Modified Files:
> > CVS:    src/examples/phidget_c.c src/examples/phidget_cpp.cc
> > CVS:    src/libphidget/phidget.c src/libphidget/phidget.h
> > CVS:    src/phidget++/CAnalogIn.cc src/phidget++/CAnalogIn.h
> > CVS:    src/phidget++/CDigitalIn.cc src/phidget++/CDigitalIn.h
> > CVS:    src/phidget++/CDigitalOut.cc src/phidget++/CDigitalOut.h
> > CVS:    src/phidget++/CInterfaceKit.cc src/phidget++/CPhidget.cc
> > CVS:    src/phidget++/CPhidget.h src/phidget++/CPhidgetManager.cc
> > CVS:    src/phidget++/CPhidgetManager.h src/phidget++/CServo.cc
> > CVS:    src/phidget++/CServo.h src/phidget++/CServoController.cc
> > CVS:    src/phidget++/CServoController.h src/phidget++/CUID.cc
> > CVS:    src/phidget++/CUID.h src/phidget++/CUniqueDevice.cc
> > CVS:    src/phidget++/CUniqueDevice.h src/phidget++/TSingleton.h
> > CVS:    src/phidget++/lsphidget.cc
> > CVS:
> > ----------------------------------------------------------------------
> > 
> > 
> > I didn't go ahead with it, but is it going to actually commit all these
> > files, or will it actually see that while the date change, nothing in
> > them changed, just wondering.
> 
> Do 'cvs diff' and see what's different.
> 
> Often, CVS goes nuts even if just the timestamp changed - it will report
> those files on commit time after time. If the 'cvs diff' is empty, you're
> fine. If it is not, run 'make indent' again couple of times and see if it
> goes away.
> 
> I'm thinking about a) looking at the parts of the code it keeps juggling b)
> submitting a bug report to indent development team. This behavior defeats
> the purpose...

"make indent" should put the code in the same form that the is stored in
CVS right?

well here is what I do after checking it out:

[jstrohm@oishi libphidget]$ bootstrap
[jstrohm@oishi libphidget]$ cvs diff
js...@cv...'s password:
cvs server: Diffing .
cvs server: Diffing misc
cvs server: Diffing src
cvs server: Diffing src/examples
cvs server: Diffing src/libphidget
cvs server: Diffing src/phidget++
cvs server: Diffing src/rpm
cvs server: Diffing src/scripts
cvs server: Diffing src/scripts/install
[jstrohm@oishi libphidget]$ cvs diff
js...@cv...'s password:
cvs server: Diffing .
cvs server: Diffing misc
cvs server: Diffing src
cvs server: Diffing src/examples
cvs server: Diffing src/libphidget
cvs server: Diffing src/phidget++
Index: src/phidget++/CAnalogIn.cc
===================================================================
RCS file: /cvsroot/libphidget/libphidget/src/phidget++/CAnalogIn.cc,v
retrieving revision 1.3
diff -u -r1.3 CAnalogIn.cc
--- src/phidget++/CAnalogIn.cc  8 Sep 2002 18:59:54 -0000       1.3
+++ src/phidget++/CAnalogIn.cc  10 Sep 2002 00:14:11 -0000
@@ -12,6 +12,8 @@

 CAnalogIn::CAnalogIn (CInterfaceKit * interfacekit, int id):
 CUniqueDevice (interfacekit, id),
-_interfaceKit (interfacekit), _value (0), _lastValue (0)
+_interfaceKit (interfacekit),
+_value (0),
+_lastValue (0)
 {
 }
Index: src/phidget++/CDigitalIn.cc
===================================================================
RCS file: /cvsroot/libphidget/libphidget/src/phidget++/CDigitalIn.cc,v
retrieving revision 1.3
diff -u -r1.3 CDigitalIn.cc
--- src/phidget++/CDigitalIn.cc 8 Sep 2002 18:59:54 -0000       1.3
+++ src/phidget++/CDigitalIn.cc 10 Sep 2002 00:14:11 -0000
@@ -11,6 +11,8 @@
 #include "CInterfaceKit.h"

 CDigitalIn::CDigitalIn (CInterfaceKit * interfacekit, int id):
-CUniqueDevice (interfacekit, id), _value (false), _lastValue (false)
+CUniqueDevice (interfacekit, id),
+_value (false),
+_lastValue (false)
 {
 }
Index: src/phidget++/CDigitalOut.cc
===================================================================
RCS file: /cvsroot/libphidget/libphidget/src/phidget++/CDigitalOut.cc,v
retrieving revision 1.3
diff -u -r1.3 CDigitalOut.cc
--- src/phidget++/CDigitalOut.cc        8 Sep 2002 18:59:54 -0000      
1.3
+++ src/phidget++/CDigitalOut.cc        10 Sep 2002 00:14:11 -0000
@@ -11,7 +11,8 @@
 #include "CInterfaceKit.h"

 CDigitalOut::CDigitalOut (CInterfaceKit * interfacekit, int id):
-CUniqueDevice (interfacekit, id), _value (false)
+CUniqueDevice (interfacekit, id),
+_value (false)
 {
 }

.
.
.
.
.
.

Re: [Libphidget-devel] just playing with indent . . .

[Vadim T. <vt...@fr...>]

According to Jack Strohm:

> Now with indent support I did this:
> 
> cvs update
> make indent-local
> make indent
> cvs commit
> 
> And I get this:
> 
> CVS:
> ----------------------------------------------------------------------
> CVS: Enter Log.  Lines beginning with `CVS:' are removed automatically
> CVS:
> CVS: Committing in .
> CVS:
> CVS: Modified Files:
> CVS:    src/examples/phidget_c.c src/examples/phidget_cpp.cc
> CVS:    src/libphidget/phidget.c src/libphidget/phidget.h
> CVS:    src/phidget++/CAnalogIn.cc src/phidget++/CAnalogIn.h
> CVS:    src/phidget++/CDigitalIn.cc src/phidget++/CDigitalIn.h
> CVS:    src/phidget++/CDigitalOut.cc src/phidget++/CDigitalOut.h
> CVS:    src/phidget++/CInterfaceKit.cc src/phidget++/CPhidget.cc
> CVS:    src/phidget++/CPhidget.h src/phidget++/CPhidgetManager.cc
> CVS:    src/phidget++/CPhidgetManager.h src/phidget++/CServo.cc
> CVS:    src/phidget++/CServo.h src/phidget++/CServoController.cc
> CVS:    src/phidget++/CServoController.h src/phidget++/CUID.cc
> CVS:    src/phidget++/CUID.h src/phidget++/CUniqueDevice.cc
> CVS:    src/phidget++/CUniqueDevice.h src/phidget++/TSingleton.h
> CVS:    src/phidget++/lsphidget.cc
> CVS:
> ----------------------------------------------------------------------
> 
> 
> I didn't go ahead with it, but is it going to actually commit all these
> files, or will it actually see that while the date change, nothing in
> them changed, just wondering.

Do 'cvs diff' and see what's different.

Often, CVS goes nuts even if just the timestamp changed - it will report
those files on commit time after time. If the 'cvs diff' is empty, you're
fine. If it is not, run 'make indent' again couple of times and see if it
goes away.

I'm thinking about a) looking at the parts of the code it keeps juggling b)
submitting a bug report to indent development team. This behavior defeats
the purpose...

--vt

[Libphidget-devel] just playing with indent . . .

[Jack S. <js...@ja...>]

Now with indent support I did this:

cvs update
make indent-local
make indent
cvs commit

And I get this:

CVS:
----------------------------------------------------------------------
CVS: Enter Log.  Lines beginning with `CVS:' are removed automatically
CVS:
CVS: Committing in .
CVS:
CVS: Modified Files:
CVS:    src/examples/phidget_c.c src/examples/phidget_cpp.cc
CVS:    src/libphidget/phidget.c src/libphidget/phidget.h
CVS:    src/phidget++/CAnalogIn.cc src/phidget++/CAnalogIn.h
CVS:    src/phidget++/CDigitalIn.cc src/phidget++/CDigitalIn.h
CVS:    src/phidget++/CDigitalOut.cc src/phidget++/CDigitalOut.h
CVS:    src/phidget++/CInterfaceKit.cc src/phidget++/CPhidget.cc
CVS:    src/phidget++/CPhidget.h src/phidget++/CPhidgetManager.cc
CVS:    src/phidget++/CPhidgetManager.h src/phidget++/CServo.cc
CVS:    src/phidget++/CServo.h src/phidget++/CServoController.cc
CVS:    src/phidget++/CServoController.h src/phidget++/CUID.cc
CVS:    src/phidget++/CUID.h src/phidget++/CUniqueDevice.cc
CVS:    src/phidget++/CUniqueDevice.h src/phidget++/TSingleton.h
CVS:    src/phidget++/lsphidget.cc
CVS:
----------------------------------------------------------------------


I didn't go ahead with it, but is it going to actually commit all these
files, or will it actually see that while the date change, nothing in
them changed, just wondering.

Re: [Libphidget-devel] bootstrap error

[Vadim T. <vt...@fr...>]

According to Jack Strohm:
> On Mon, 2002-09-09 at 18:44, Vadim Tkachenko wrote:
> > According to Jack Strohm:
> > 
> > > I get this when I bootstrap, think you didn't add ./src/rpm/Makefile.in
> > > and ./src/rpm/libphidget.spec.in
> > > 
> > > [jstrohm@oishi libphidget]$ bootstrap
> > > src/Makefile.am:1: required directory src/rpm does not exist
> > 
> > I think you're having a problem with your CVS setup - take a look at
> > ./src/rpm directory, it is most probably empty.
> > 
> > Now, try to create $HOME/.cvsrc file and write the following into it:
> > 
> > diff -u
> > cvs -z9
> > update -d -P
> > 
> > Then do 'cvs update' from the top level again, and I bet the files will
> > appear, and everything will be fine.
> > 
> > Let me know if this works.
> > 
> 
> that worked, what do those options mean?

Short answer: RTFM :)

Long answer: diff and cvs option don't mean anything in this case (the way
the diff looks and the compression ratio), but update
-d and update -P do.
 
From 'man cvs':

Use the -d option to create any directories that exist in the repository if
they're missing from the working directory.  (Normally, update acts only on
directories and files that were already enrolled in your working directory.)
This is useful for updating directories that were created in the repository
since the initial checkout; but it has an unfortunate side effect.  If you
deliberately avoided certain directories in the repository when you created
your working directory (either through use of a module name or by listing
explicitly the files and directories you wanted on the command line), then
updating with -d will create those directories, which may not be what you
want.

-P     Prune  (remove)  directories that are empty after being
       updated, on checkout, or update.  Normally, an empty directory (one
       that is void of revision-controlled files) is left alone. Specifying
       -P will cause these directories to be silently removed from your
       checked-out sources.  This does not remove the directory from the
       repository, only from your checked out copy.  Note that this option
       is implied by the -r or -D options of checkout and export.

Note that these are context-dependent, -d will mean absolutely different
things in commands other than 'update'.

--vt

Re: [Libphidget-devel] bootstrap error

[Jack S. <js...@ja...>]

On Mon, 2002-09-09 at 18:44, Vadim Tkachenko wrote:
> According to Jack Strohm:
> 
> > I get this when I bootstrap, think you didn't add ./src/rpm/Makefile.in
> > and ./src/rpm/libphidget.spec.in
> > 
> > [jstrohm@oishi libphidget]$ bootstrap
> > src/Makefile.am:1: required directory src/rpm does not exist
> 
> I think you're having a problem with your CVS setup - take a look at
> ./src/rpm directory, it is most probably empty.
> 
> Now, try to create $HOME/.cvsrc file and write the following into it:
> 
> diff -u
> cvs -z9
> update -d -P
> 
> Then do 'cvs update' from the top level again, and I bet the files will
> appear, and everything will be fine.
> 
> Let me know if this works.
> 

that worked, what do those options mean?

Re: [Libphidget-devel] bootstrap error

[Vadim T. <vt...@fr...>]

According to Jack Strohm:

> I get this when I bootstrap, think you didn't add ./src/rpm/Makefile.in
> and ./src/rpm/libphidget.spec.in
> 
> [jstrohm@oishi libphidget]$ bootstrap
> src/Makefile.am:1: required directory src/rpm does not exist

I think you're having a problem with your CVS setup - take a look at
./src/rpm directory, it is most probably empty.

Now, try to create $HOME/.cvsrc file and write the following into it:

diff -u
cvs -z9
update -d -P

Then do 'cvs update' from the top level again, and I bet the files will
appear, and everything will be fine.

Let me know if this works.

--vt

[Libphidget-devel] bootstrap error

[Jack S. <js...@ja...>]

I get this when I bootstrap, think you didn't add ./src/rpm/Makefile.in
and ./src/rpm/libphidget.spec.in

[jstrohm@oishi libphidget]$ bootstrap
src/Makefile.am:1: required directory src/rpm does not exist

Re: [Libphidget-devel] RPM build: sketch done

[Jack S. <js...@ja...>]

On Mon, 2002-09-09 at 18:04, Vadim Tkachenko wrote:
> Hello,
> 
> Do 'cvs update', become root and do 'make rpm' - it'll create the binary and
> source RPMs at the proper locations.
> 
> Restrictions:
> 
> - This has been tested on RedHat 7.0, no idea how will it work on other
>   distributions;

I'll test it n my box.  I got a redhat 7.3 at work and a mandrake 9beta3

> 
> - Documentation included with the RPM package is by no means adequate. I was
>   thinking about adding the files generated by Doxygen to the package (as
>   well as to 'make dist'), but I don't want to shuffle them around myself.
>   Proposal is to create ${srcdir}/docs and generate into that directory, and
>   later deploy as part of 'make install'. Another consideration - well, may
>   be it *is* a good idea to make the website a part of the distribution?
>   Then, you'd have consistent site navigation all around...

hmmm . . .


> 
> - Can't make rpm install the package without --nodeps - complains about
>   missing libusb (right, I don't have the *package* installed) and libsafe
>   (beats me).
> 

yeah, no libusb rpm available yet.


as for libsafe:

http://rpmfind.net/linux/RPM/powertools/6.2/i386/i386/libsafe-1.1-1.i386.html

that is weird, wonder why it wants that?

Re: [Libphidget-devel] Teaser ;)

[Vadim T. <vt...@fr...>]

According to Jack Strohm:

> you said you had to be root to run make rpm, why is that.  Is it making
> and installing an RPM?

Making, but not installing.

Take a look at the bottom of the toplevel Makefile.am - there are things
that *absolutely* must be done in order for RPM to work, and they include
modifying the content of /usr/src/redhat - can't be done without being root.

I figured that it's not really necessary to make the make make ;) the RPM
and install it too - it's rather a distribution thing. You don't want to
install the RPM as a part of a debugging process - it includes complete
recompilation everytime and is quite slow. 'make install' will suffice, but
RPM package is a neat thing when it comes to making the user happy ;)

--vt

[Libphidget-devel] RPM build: sketch done

[Vadim T. <vt...@fr...>]

Hello,

Do 'cvs update', become root and do 'make rpm' - it'll create the binary and
source RPMs at the proper locations.

Restrictions:

- This has been tested on RedHat 7.0, no idea how will it work on other
  distributions;

- Documentation included with the RPM package is by no means adequate. I was
  thinking about adding the files generated by Doxygen to the package (as
  well as to 'make dist'), but I don't want to shuffle them around myself.
  Proposal is to create ${srcdir}/docs and generate into that directory, and
  later deploy as part of 'make install'. Another consideration - well, may
  be it *is* a good idea to make the website a part of the distribution?
  Then, you'd have consistent site navigation all around...

- Can't make rpm install the package without --nodeps - complains about
  missing libusb (right, I don't have the *package* installed) and libsafe
  (beats me).

Other than that, works just fine...

--vt

Re: [Libphidget-devel] Teaser ;)

[Jack S. <js...@ja...>]

On Mon, 2002-09-09 at 17:56, Jack Strohm wrote:
> On Mon, 2002-09-09 at 17:15, Vadim Tkachenko wrote:
> > Hello,
> > 
> > -----------------8<-----------------------------------------------
> > 15:11:42 root@freehold:/usr/src/redhat/RPMS/i386$ rpm -Uvh --nodeps libphidget-0.1p0-1.i386.rpm 
> > Preparing...                ########################################### [100%]
> >    1:libphidget             ########################################### [100%]
> > 
> > 15:12:03 root@freehold:/usr/src/redhat/RPMS/i386$ rpm -qi libphidget                  
> > Name        : libphidget                   Relocations: (not relocateable)
> > Version     : 0.1p0                             Vendor: (none)
> > Release     : 1                             Build Date: Mon 09 Sep 2002 03:09:18 PM MST
> > Install date: Mon 09 Sep 2002 03:11:52 PM MST      Build Host: freehold.crocodile.org
> > Group       : Development/Tools             Source RPM: libphidget-0.1p0-1.src.rpm
> > Size        : 617820                           License: LGPL
> > Summary     : Phidget Library
> > Description :
> > Phidget Library provides C/C++ library to operate the Phidget family of
> > devices (http://www.phidgets.com/).
> > -----------------8<-----------------------------------------------
> > 
> > Doesn't mean it works, though ;) Figuring out how to do the symlinks.
> > 
> 
> cool!
> 
> good luck!
> 

you said you had to be root to run make rpm, why is that.  Is it making
and installing an RPM?

Re: [Libphidget-devel] Teaser ;)

[Jack S. <js...@ja...>]

On Mon, 2002-09-09 at 17:15, Vadim Tkachenko wrote:
> Hello,
> 
> -----------------8<-----------------------------------------------
> 15:11:42 root@freehold:/usr/src/redhat/RPMS/i386$ rpm -Uvh --nodeps libphidget-0.1p0-1.i386.rpm 
> Preparing...                ########################################### [100%]
>    1:libphidget             ########################################### [100%]
> 
> 15:12:03 root@freehold:/usr/src/redhat/RPMS/i386$ rpm -qi libphidget                  
> Name        : libphidget                   Relocations: (not relocateable)
> Version     : 0.1p0                             Vendor: (none)
> Release     : 1                             Build Date: Mon 09 Sep 2002 03:09:18 PM MST
> Install date: Mon 09 Sep 2002 03:11:52 PM MST      Build Host: freehold.crocodile.org
> Group       : Development/Tools             Source RPM: libphidget-0.1p0-1.src.rpm
> Size        : 617820                           License: LGPL
> Summary     : Phidget Library
> Description :
> Phidget Library provides C/C++ library to operate the Phidget family of
> devices (http://www.phidgets.com/).
> -----------------8<-----------------------------------------------
> 
> Doesn't mean it works, though ;) Figuring out how to do the symlinks.
> 

cool!

good luck!

[Libphidget-devel] Teaser ;)

[Vadim T. <vt...@fr...>]

Hello,

-----------------8<-----------------------------------------------
15:11:42 root@freehold:/usr/src/redhat/RPMS/i386$ rpm -Uvh --nodeps libphidget-0.1p0-1.i386.rpm 
Preparing...                ########################################### [100%]
   1:libphidget             ########################################### [100%]

15:12:03 root@freehold:/usr/src/redhat/RPMS/i386$ rpm -qi libphidget                  
Name        : libphidget                   Relocations: (not relocateable)
Version     : 0.1p0                             Vendor: (none)
Release     : 1                             Build Date: Mon 09 Sep 2002 03:09:18 PM MST
Install date: Mon 09 Sep 2002 03:11:52 PM MST      Build Host: freehold.crocodile.org
Group       : Development/Tools             Source RPM: libphidget-0.1p0-1.src.rpm
Size        : 617820                           License: LGPL
Summary     : Phidget Library
Description :
Phidget Library provides C/C++ library to operate the Phidget family of
devices (http://www.phidgets.com/).
-----------------8<-----------------------------------------------

Doesn't mean it works, though ;) Figuring out how to do the symlinks.

--vt

Re: [Libphidget-devel] Deployment layout

[Jack S. <js...@ja...>]

On Mon, 2002-09-09 at 05:07, Vadim Tkachenko wrote:
> Hello,
> 
> Which include files have to be exposed to the library users? OK, from the C
> side it's phidget.h, what about C++? Some of them are private, right?

I'll have to get back to you on that.  I think they all "might" be
needed, depending on if you create the object or not.

> 
> Does it make sense to separate the C headers from C++ headers at the
> deployment location (--prefix)? My guess is no.

at deployment location, no they can go to the samelocation.