GRASS 5.0.3 release candidate 4 (RC4)
... is available for testing:
http://mpa.itc.it/markus/grass50/
TODO: - update NEWS.html for the release
- else?
Markus
On Thu, Oct 09, 2003 at 11:42:43AM +0200, Markus Neteler wrote:
GRASS 5.0.3 release candidate 4 (RC4)
... is available for testing:
http://mpa.itc.it/markus/grass50/
TODO: - update NEWS.html for the release
- else?
* The automatically created Changelog is missing the removal
of the apngs code (because this was no commit).
I suggest adding a hint manually.
* Unfortunately src.nonGPL got partly packaged,
this needs to be removed from the tarball.
* I'm unsure if we should package all code that actually is not used
in the tarball. We still have it in the CVS.
Like src.garden/grass.java, unused, src.todo and some other stuff
from src.garden and src.contrib.
I suggest to not package them.
* It would be good style to create a detached signature for the
tarball in addition to the useful md5sums.
I suggest using Gnupg and a key of Markus as packager for the
time being.
On Thu, Oct 09, 2003 at 12:54:43PM +0200, Bernhard Reiter wrote:
On Thu, Oct 09, 2003 at 11:42:43AM +0200, Markus Neteler wrote:
> GRASS 5.0.3 release candidate 4 (RC4)
>
> ... is available for testing:
>
> http://mpa.itc.it/markus/grass50/
>
> TODO: - update NEWS.html for the release
> - else?* The automatically created Changelog is missing the removal
of the apngs code (because this was no commit).
I suggest adding a hint manually.
OK, we can do that but it will disappear in the next Changelog.
Better to use a commit entry.
* Unfortunately src.nonGPL got partly packaged,
this needs to be removed from the tarball.
Strange:
cat src.nonGPL/CVS/Tag
Treleasebranch_26_april_2002_5_0_0
Why is that tagged? Did you tag it (please remove then)?
I have just updated the tree.
* I'm unsure if we should package all code that actually is not used
in the tarball. We still have it in the CVS.
Like src.garden/grass.java, unused, src.todo and some other stuff
from src.garden and src.contrib.
I suggest to not package them.
It it should be removed from the release branch.
* It would be good style to create a detached signature for the
tarball in addition to the useful md5sums.
I suggest using Gnupg and a key of Markus as packager for the
time being.
I have no idea how to do that - hints are welcome.
Markus
Markus Neteler wrote:
> * Unfortunately src.nonGPL got partly packaged,
> this needs to be removed from the tarball.Strange:
cat src.nonGPL/CVS/Tag
Treleasebranch_26_april_2002_5_0_0Why is that tagged? Did you tag it (please remove then)?
I have just updated the tree.
src.nonGPL wasn't removed until after the release branch was created
(AFAICT, it was removed the following day).
Did you use "cvs update -dP"? Empty directories will only be removed
if you use the "-P" switch. Similarly, directories which exist in the
repository but not in your local copy will only be added if you use
the "-d" switch.
If you checked out a copy of the release branch on the day it was
created, then updated it without "-P" (or if you made a copy of the
head then updated it to the release branch without "-P"), the
src.nonGPL directory (and its subdirectories) will still exist (with
the appropriate tag), they just won't contain any files.
Personally, I have "update -dP" in my ~/.cvsrc file, so "cvs update"
will always add and remove directories as necessary.
--
Glynn Clements <glynn.clements@virgin.net>
On Thu, Oct 09, 2003 at 02:01:06PM +0100, Glynn Clements wrote:
Markus Neteler wrote:
> > * Unfortunately src.nonGPL got partly packaged,
> > this needs to be removed from the tarball.
>
> Strange:
> cat src.nonGPL/CVS/Tag
> Treleasebranch_26_april_2002_5_0_0
>
> Why is that tagged? Did you tag it (please remove then)?
> I have just updated the tree.src.nonGPL wasn't removed until after the release branch was created
(AFAICT, it was removed the following day).Did you use "cvs update -dP"? Empty directories will only be removed
if you use the "-P" switch. Similarly, directories which exist in the
repository but not in your local copy will only be added if you use
the "-d" switch.
It also comes when using "cvs up -dP".
The problem is that r.agnps50 was already tagged, now the tags are
moved as well.
If you checked out a copy of the release branch on the day it was
created, then updated it without "-P" (or if you made a copy of the
head then updated it to the release branch without "-P"), the
src.nonGPL directory (and its subdirectories) will still exist (with
the appropriate tag), they just won't contain any files.
Does this apply for files moved in CVS server?
Should it be just removed from the branch?
Personally, I have "update -dP" in my ~/.cvsrc file, so "cvs update"
will always add and remove directories as necessary.
Good idea!
Markus
On Thu, Oct 09, 2003 at 01:18:18PM +0200, Markus Neteler wrote:
On Thu, Oct 09, 2003 at 12:54:43PM +0200, Bernhard Reiter wrote:
> On Thu, Oct 09, 2003 at 11:42:43AM +0200, Markus Neteler wrote:
> > GRASS 5.0.3 release candidate 4 (RC4)
> >
> > ... is available for testing:
> >
> > http://mpa.itc.it/markus/grass50/
> >
> > TODO: - update NEWS.html for the release
> > - else?
>
> * The automatically created Changelog is missing the removal
> of the apngs code (because this was no commit).
> I suggest adding a hint manually.OK, we can do that but it will disappear in the next Changelog.
Better to use a commit entry.
I'd say in general that it is _not_ a good idea to
let the cvs2changelog script overwrite old entries.
The better changelogs I know are at least controlled manually.
> * Unfortunately src.nonGPL got partly packaged,
> this needs to be removed from the tarball.Strange:
cat src.nonGPL/CVS/Tag
Treleasebranch_26_april_2002_5_0_0Why is that tagged? Did you tag it (please remove then)?
I have just updated the tree.
Yes I haven't thought of that the tag got moved, too.
Also I believed that your packaging script might be actively
selecting the needed source, not just packaging everything. 
Anyway I've removed the subdirectories of src.nonGPL from the branch now.
> * I'm unsure if we should package all code that actually is not used
> in the tarball. We still have it in the CVS.
> Like src.garden/grass.java, unused, src.todo and some other stuff
> from src.garden and src.contrib.
> I suggest to not package them.It it should be removed from the release branch.
Do you want to do this (for exercise)?
It is easy enough:
If you have a fully checked out branch, just
use something like
cd unused
cvs remove -f
cvs commit
> * It would be good style to create a detached signature for the
> tarball in addition to the useful md5sums.
> I suggest using Gnupg and a key of Markus as packager for the
> time being.I have no idea how to do that - hints are welcome.
www.gnupg.org has good documentation for gnupg.
Steps (briefly):
Install an uptodate version of gpg.
Create a key if you don't have one:
http://www.gnupg.org/(de)/howtos/de/GPGMiniHowto-3.html#ss3.1
gpg --gen-key
(1024 bit is sufficient.)
Sign a file with a detached signature
gpg --detach-sign grass-5.0.3rc4_src.tar.gz
( Later and repeately:
Get the public key part signed by people to build
the web of trust. )
On Thu, Oct 09, 2003 at 04:08:42PM +0200, Bernhard Reiter wrote:
On Thu, Oct 09, 2003 at 01:18:18PM +0200, Markus Neteler wrote:
> On Thu, Oct 09, 2003 at 12:54:43PM +0200, Bernhard Reiter wrote:
> > On Thu, Oct 09, 2003 at 11:42:43AM +0200, Markus Neteler wrote:
> > > GRASS 5.0.3 release candidate 4 (RC4)
> > >
> > > ... is available for testing:
> > >
> > > http://mpa.itc.it/markus/grass50/
> > >
> > > TODO: - update NEWS.html for the release
> > > - else?
> >
> > * The automatically created Changelog is missing the removal
> > of the apngs code (because this was no commit).
> > I suggest adding a hint manually.
>
> OK, we can do that but it will disappear in the next Changelog.
> Better to use a commit entry.I'd say in general that it is _not_ a good idea to
let the cvs2changelog script overwrite old entries.
The better changelogs I know are at least controlled manually.
So you are welcome to do this job I won't do that.
Since 1/2000 the Changlog is created like this, so we
may live with it further on. Note that there is also
NEWS.html which is maintained manually (developers may add
missing entries directly!).
> > * Unfortunately src.nonGPL got partly packaged,
> > this needs to be removed from the tarball.
>
> Strange:
> cat src.nonGPL/CVS/Tag
> Treleasebranch_26_april_2002_5_0_0
>
> Why is that tagged? Did you tag it (please remove then)?
> I have just updated the tree.Yes I haven't thought of that the tag got moved, too.
Also I believed that your packaging script might be actively
selecting the needed source, not just packaging everything.
I am extracting from CVS, adding Changelog (via cvs2cl.pl),
making a tar ball, that's it more or less. There is no magic inside.
If things like src.nonGPL/ are inside, it's a problem of the
release_branch and not an error in packaging.
I think that we want have everything reproducable. So we have to
maintain a *clean* release_branch.
Anyway I've removed the subdirectories of src.nonGPL from the branch now.
OK, now it's gone.
> > * I'm unsure if we should package all code that actually is not used
> > in the tarball. We still have it in the CVS.
> > Like src.garden/grass.java, unused, src.todo and some other stuff
> > from src.garden and src.contrib.
> > I suggest to not package them.
>
> It it should be removed from the release branch.Do you want to do this (for exercise)?
It is easy enough:
If you have a fully checked out branch, just
use something like
cd unused
cvs remove -f
cvs commit
That's fun
OK, removed
src.todo/
src.garden/grass.hdf/
src.garden/grass.java/
> > * It would be good style to create a detached signature for the
> > tarball in addition to the useful md5sums.
> > I suggest using Gnupg and a key of Markus as packager for the
> > time being.
>
> I have no idea how to do that - hints are welcome.www.gnupg.org has good documentation for gnupg.
Steps (briefly):
Install an uptodate version of gpg.Create a key if you don't have one:
http://www.gnupg.org/(de)/howtos/de/GPGMiniHowto-3.html#ss3.1
gpg --gen-key
(1024 bit is sufficient.)Sign a file with a detached signature
gpg --detach-sign grass-5.0.3rc4_src.tar.gz( Later and repeately:
Get the public key part signed by people to build
the web of trust.
I'll see if I get this working (too many things at the same time).
Markus
On Thu, Oct 09, 2003 at 04:36:08PM +0200, Markus Neteler wrote:
On Thu, Oct 09, 2003 at 04:08:42PM +0200, Bernhard Reiter wrote:
> On Thu, Oct 09, 2003 at 01:18:18PM +0200, Markus Neteler wrote:
> > On Thu, Oct 09, 2003 at 12:54:43PM +0200, Bernhard Reiter wrote:
> > > On Thu, Oct 09, 2003 at 11:42:43AM +0200, Markus Neteler wrote:
> > > > GRASS 5.0.3 release candidate 4 (RC4)
> > > >
> > > > ... is available for testing:
> > > >
> > > > http://mpa.itc.it/markus/grass50/
> > > >
> > > > TODO: - update NEWS.html for the release
> > > > - else?
> > >
> > > * The automatically created Changelog is missing the removal
> > > of the apngs code (because this was no commit).
> > > I suggest adding a hint manually.
> >
> > OK, we can do that but it will disappear in the next Changelog.
> > Better to use a commit entry.
>
> I'd say in general that it is _not_ a good idea to
> let the cvs2changelog script overwrite old entries.
> The better changelogs I know are at least controlled manually.So you are welcome to do this job
Since 1/2000 the Changlog is created like this, so we
may live with it further on. Note that there is also
NEWS.html which is maintained manually (developers may add
missing entries directly!).
True, NEWS.html will be sufficient for the time being I guess.
> Also I believed that your packaging script might be actively
> selecting the needed source, not just packaging everything.I am extracting from CVS, adding Changelog (via cvs2cl.pl),
making a tar ball, that's it more or less. There is no magic inside.
If things like src.nonGPL/ are inside, it's a problem of the
release_branch and not an error in packaging.
Yes, it was my missconception of how it would work.
Anyway we are constantly improving, I would say. 
> > > * I'm unsure if we should package all code that actually is not used
> > > in the tarball. We still have it in the CVS.
> > > Like src.garden/grass.java, unused, src.todo and some other stuff
> > > from src.garden and src.contrib.
> > > I suggest to not package them.
> >
> > It it should be removed from the release branch.
>
> Do you want to do this (for exercise)?
>
> It is easy enough:
> If you have a fully checked out branch, just
> use something like
> cd unused
> cvs remove -f
> cvs commitThat's fun
OK, removedsrc.todo/
src.garden/grass.hdf/
src.garden/grass.java/
What about "unused"?
and informix and oracle2?
On Thu, Oct 09, 2003 at 05:07:34PM +0200, Bernhard Reiter wrote:
On Thu, Oct 09, 2003 at 04:36:08PM +0200, Markus Neteler wrote:
> On Thu, Oct 09, 2003 at 04:08:42PM +0200, Bernhard Reiter wrote:
> > On Thu, Oct 09, 2003 at 01:18:18PM +0200, Markus Neteler wrote:
> > > On Thu, Oct 09, 2003 at 12:54:43PM +0200, Bernhard Reiter wrote:
> > > > On Thu, Oct 09, 2003 at 11:42:43AM +0200, Markus Neteler wrote:
> > > > > GRASS 5.0.3 release candidate 4 (RC4)
[...]
> > It is easy enough:
> > If you have a fully checked out branch, just
> > use something like
> > cd unused
> > cvs remove -f
> > cvs commit
>
> That's fun
> OK, removed
>
> src.todo/
> src.garden/grass.hdf/
> src.garden/grass.java/What about "unused"?
Where is that?
and informix and oracle2?
Removed.
I think it's ok now, we don't have to search for a few bytes
to save. Let's finish this endless story (we started in August)
and work on 5.3/5.7. Most users will await a 5.3.0 release to
get datum transformation support and the NVIZ updates...
Markus
On Thu, Oct 09, 2003 at 05:42:36PM +0200, Markus Neteler wrote:
On Thu, Oct 09, 2003 at 05:07:34PM +0200, Bernhard Reiter wrote:
> What about "unused"?
Where is that?
tar -tvmlzf /download/src/grass-5.0.3rc4_src.tar.gz | grep /unused | tail -1
-rw-r--r-- neteler/ssi 31158 2000-11-11 01:03:44 grass-5.0.3rc4/unused/tcltkgrass/src/view.c
I think it's ok now, we don't have to search for a few bytes
to save.
Yes, I was just wondering.
Let's finish this endless story (we started in August)
and work on 5.3/5.7. Most users will await a 5.3.0 release to
get datum transformation support and the NVIZ updates...
Yes!
Though we might provide _some_ support for 5.0.x as long
as 5.4.x is not out. Only _grave_ _critical_ bugs of course.
Let's go towards 5.4 where 5.3.0 is the next step.
On Thu, Oct 09, 2003 at 06:10:24PM +0200, Bernhard Reiter wrote:
On Thu, Oct 09, 2003 at 05:42:36PM +0200, Markus Neteler wrote:
> On Thu, Oct 09, 2003 at 05:07:34PM +0200, Bernhard Reiter wrote:> > What about "unused"?
> Where is that?tar -tvmlzf /download/src/grass-5.0.3rc4_src.tar.gz | grep /unused | tail -1
-rw-r--r-- neteler/ssi 31158 2000-11-11 01:03:44 grass-5.0.3rc4/unused/tcltkgrass/src/view.c> I think it's ok now, we don't have to search for a few bytes
> to save.Yes, I was just wondering.
Sorry, found unused/ (was looking at the dir with ls -lart).
Deleted now.
> Let's finish this endless story (we started in August)
> and work on 5.3/5.7. Most users will await a 5.3.0 release to
> get datum transformation support and the NVIZ updates...Yes!
Though we might provide _some_ support for 5.0.x as long
as 5.4.x is not out. Only _grave_ _critical_ bugs of course.
Right.
Let's go towards 5.4 where 5.3.0 is the next step.
Tomorrow I'll package RC5.
Markus
On Thu, Oct 09, 2003 at 08:20:26PM +0200, Markus Neteler wrote:
Tomorrow I'll package RC5.
Cool!
> * I'm unsure if we should package all code that actually is not used
> in the tarball. We still have it in the CVS.
> Like src.garden/grass.java, unused, src.todo and some other stuff
> from src.garden and src.contrib.
> I suggest to not package them.It it should be removed from the release branch.
I'd argue for keeping it in, otherwise new/experimental/inspiration code
never gets tried by anyone but the very adventurous. Having it CVS-only
puts the barrier too high for beginners at GRASS coding. It will only
exist for those compiling from source, the sort of person who'd probably
end up using 'gmake5' for little things anyway. The binaries won't
change size. Let the 5.7 cvs sort out the old & unused code from the
still experimental; pretty much everything in the 5.0.x branch applies
to 5.0.x, no? Is there lots of unported 4.3 left?
Certaintly leave everything in the 5.3 snapshots.
2c,
Hamish
Though we might provide _some_ support for 5.0.x as long
as 5.4.x is not out. Only _grave_ _critical_ bugs of course.
Would that mean grave & critical[*] bugs in *specific modules* or
in the *overall system* (libraries & init.sh/gisenv stuff)?
[*] I take those to mean:
+ non-functional (SegFaults while not doing something weird)
+ data corruption of existing map
+ significantly erroneous data output
+ license issue
+ anything else?
?
Hamish
GRASS 5.0.3 release candidate 4 (RC4)
r.terraflow should either be updated to the latest or left out.
yes:
src/display/d.text.freetype/main.c
minor:
src/sites/s.random/main.c
html/html/s.random.html
src/tcltkgrass/module/r.what
src/tcltkgrass/module/d.what.rast
src/tcltkgrass/module/v.what
src/tcltkgrass/module/v.what
src/tcltkgrass/module/d.what.vect
src/display/d.text.freetype/test.input
html/html/d.text.freetype.html
src/display/d.label/cmd/label.c
src/display/d.title/cmd/main.c
html/html/d.title.html
src/display/d.geodesic/cmd/main.c
INSTALL
thanks,
Hamish
On Fri, Oct 10, 2003 at 09:08:32PM +1300, Hamish wrote:
> > * I'm unsure if we should package all code that actually is not used
> > in the tarball. We still have it in the CVS.
> > Like src.garden/grass.java, unused, src.todo and some other stuff
> > from src.garden and src.contrib.
> > I suggest to not package them.
>
> It it should be removed from the release branch.I'd argue for keeping it in, otherwise new/experimental/inspiration code
never gets tried by anyone but the very adventurous. Having it CVS-only
puts the barrier too high for beginners at GRASS coding. It will only
exist for those compiling from source, the sort of person who'd probably
end up using 'gmake5' for little things anyway. The binaries won't
change size. Let the 5.7 cvs sort out the old & unused code from the
still experimental; pretty much everything in the 5.0.x branch applies
to 5.0.x, no? Is there lots of unported 4.3 left?Certaintly leave everything in the 5.3 snapshots.
The situation with most of the mentioned code above is
that most of it has problems that will significantly confuse beginners.
They might not be updated or even have subtle licensing issues.
If we want source code for GRASS developers or beginners to play with,
we should make "experimental" add-on distributions instead I think.
On Fri, Oct 10, 2003 at 10:36:27PM +1300, Hamish wrote:
> GRASS 5.0.3 release candidate 4 (RC4)
r.terraflow should either be updated to the latest or left out.
OK, r.terraflow is removed from the release_branch as discussed
earlier. It comes with 5.3-cvs onwards (have added it also to 5.7).
yes:
src/display/d.text.freetype/main.cminor:
src/sites/s.random/main.c
html/html/s.random.html
src/tcltkgrass/module/r.what
src/tcltkgrass/module/d.what.rast
src/tcltkgrass/module/v.what
src/tcltkgrass/module/v.what
src/tcltkgrass/module/d.what.vect
src/display/d.text.freetype/test.input
html/html/d.text.freetype.html
src/display/d.label/cmd/label.c
src/display/d.title/cmd/main.c
html/html/d.title.html
src/display/d.geodesic/cmd/main.c
INSTALL
Please make updates if desired. I'll wait with RC5 then.
Markus
On Fri, Oct 10, 2003 at 10:36:27PM +1300, Hamish wrote:
> GRASS 5.0.3 release candidate 4 (RC4)
r.terraflow should either be updated to the latest or left out.
yes:
src/display/d.text.freetype/main.cminor:
src/sites/s.random/main.c
html/html/s.random.html
What does "yes" and "minor" mean here?
> [important fix]:
> src/display/d.text.freetype/main.c
>
> [minor fix]:
> src/sites/s.random/main.c
> html/html/s.random.html
> src/tcltkgrass/module/r.what
> src/tcltkgrass/module/d.what.rast
> src/tcltkgrass/module/v.what
> src/tcltkgrass/module/d.what.vect
> src/display/d.text.freetype/test.input
> html/html/d.text.freetype.html
> src/display/d.label/cmd/label.c
> src/display/d.title/cmd/main.c
> html/html/d.title.html
> src/display/d.geodesic/cmd/main.c
> INSTALLPlease make updates if desired. I'll wait with RC5 then.
done.
Hamish