[GRASS-dev] html error (access file not found) on compilation of i.landsat.toar in v7

GRASS GRASS developer
1

Hello,

I have been able to compile all the i.landsat.toar C code so far in
v7, but it does something strange with the html (old description.html
renamed i.landsat.toar.html in v7).

-----------
access: No such file or directory
ERROR: LOCATION <<
       /home/icwater/grass_trunk/dist.x86_64-unknown-linux-gnu/demolocation
       >> not available
make: *** [i.landsat.toar.tmp.html] Error 1
rm i.landsat.toar.tmp.html
-----------

What is the reason?
Thank you,
Yann

--
Yann Chemin
Senior Spatial Hydrologist
www.csu.edu.au/research/icwater
M +61-4-3740 7019

2

Answering to it, it seems the error does not forbid the code to work
as a module but the manual section is not existing because of that
error.

http://trac.osgeo.org/grass/ticket/1068 mentions about the
access/LOCATION error part.

I should have a look at the html file again, maybe some formating has
changed from v6 to v7.

On 14 October 2010 14:46, Yann Chemin <yann.chemin@gmail.com> wrote:

Hello,

I have been able to compile all the i.landsat.toar C code so far in
v7, but it does something strange with the html (old description.html
renamed i.landsat.toar.html in v7).

-----------
access: No such file or directory
ERROR: LOCATION <<
/home/icwater/grass_trunk/dist.x86_64-unknown-linux-gnu/demolocation
>> not available
make: *** [i.landsat.toar.tmp.html] Error 1
rm i.landsat.toar.tmp.html
-----------

What is the reason?
Thank you,
Yann

--
Yann Chemin
Senior Spatial Hydrologist
www.csu.edu.au/research/icwater
M +61-4-3740 7019

--
Yann Chemin
Senior Spatial Hydrologist
www.csu.edu.au/research/icwater
M +61-4-3740 7019

3

Yann Chemin wrote:

ERROR: LOCATION <<
       /home/icwater/grass_trunk/dist.x86_64-unknown-linux-gnu/demolocation
       >> not available

What is the reason?

I suspect that the installed $GISBASE/demolocation/.grassrc70 file has
GISDBASE pointing to the source tree rather than to the installed
path.

I think that the real-install target in Install.make needs something
like:

  -sed 's#$(GISBASE)#$(INST_DIR)#g' \
    $(GISBASE)/demolocation/.grassrc$(GRASS_VERSION_MAJOR)$(GRASS_VERSION_MINOR) \
    > $(INST_DIR)/demolocation/.grassrc$(GRASS_VERSION_MAJOR)$(GRASS_VERSION_MINOR)

Similar to what is already done for etc/fontcap. We should probably be
doing something similar for Platform.make, although that can be
overridden on the command line when running make.

--
Glynn Clements <glynn@gclements.plus.com>

4

Glynn Clements wrote:

I think that the real-install target in Install.make needs something
like:

  -sed 's#$(GISBASE)#$(INST_DIR)#g' \
    $(GISBASE)/demolocation/.grassrc$(GRASS_VERSION_MAJOR)$(GRASS_VERSION_MINOR) \
    > $(INST_DIR)/demolocation/.grassrc$(GRASS_VERSION_MAJOR)$(GRASS_VERSION_MINOR)

Similar to what is already done for etc/fontcap. We should probably be
doing something similar for Platform.make, although that can be
overridden on the command line when running make.

FWIW, this should be fixed in r43905. The only remaining references to
the source directory should be GRASS_HOME, config.status, comments in
the ctypes wrappers and debug info in binaries.

Actually, you also need r43906, otherwise the g.mapset documentation
uses <srcdir>/dist.<arch>/demolocation as the default database and
location.

Note to developers: *don't* use context-sensitive values for option
defaults unless you have some plan to ensure that the documentation
generated by --html-description isn't bogus.

--
Glynn Clements <glynn@gclements.plus.com>

5

Glynn wrote:

> -sed 's#$(GISBASE)#$(INST_DIR)#g'\
> > $(GISBASE)/demolocation

...

fwiw, we should rename "demolocation" to "dummylocation" or
similar, as it better describes what it is and why it's there.

Hamish

6

Thank you Glynn that seems to be resolved,
A last error comes up now:

------------------------

root@SeeePC:/home/yann/grass_yann/imagery/i.landsat.toar# make
VERSION_NUMBER=7.0.svn
/home/yann/grass_yann/dist.i686-pc-linux-gnu/tools/g.html2man.py
/home/yann/grass_yann/dist.i686-pc-linux-gnu/docs/html/i.landsat.toar.html
/home/yann/grass_yann/dist.i686-pc-linux-gnu/man/man1/i.landsat.toar.1
/home/yann/grass_yann/dist.i686-pc-linux-gnu/docs/html/i.landsat.toar.html:133:0:
Error (IndexError('pop from empty list',)): </p>

make: *** [/home/yann/grass_yann/dist.i686-pc-linux-gnu/man/man1/i.landsat.toar.1]
Error 1

------------------------

The html file was OK in v6, so I suppose something new came up in that
part too. I tried to look in the file, but nothing seems to flash.

On 14 October 2010 21:05, Glynn Clements <glynn@gclements.plus.com> wrote:

Glynn Clements wrote:

I think that the real-install target in Install.make needs something
like:

  \-sed &#39;s\#$\(GISBASE\)\#$\(INST\_DIR\)\#g&#39; \\
          $\(GISBASE\)/demolocation/\.grassrc$\(GRASS\_VERSION\_MAJOR\)$\(GRASS\_VERSION\_MINOR\) \\
          &gt; $\(INST\_DIR\)/demolocation/\.grassrc$\(GRASS\_VERSION\_MAJOR\)$\(GRASS\_VERSION\_MINOR\)

Similar to what is already done for etc/fontcap. We should probably be
doing something similar for Platform.make, although that can be
overridden on the command line when running make.

FWIW, this should be fixed in r43905. The only remaining references to
the source directory should be GRASS_HOME, config.status, comments in
the ctypes wrappers and debug info in binaries.

Actually, you also need r43906, otherwise the g.mapset documentation
uses <srcdir>/dist.<arch>/demolocation as the default database and
location.

Note to developers: *don't* use context-sensitive values for option
defaults unless you have some plan to ensure that the documentation
generated by --html-description isn't bogus.

--
Glynn Clements <glynn@gclements.plus.com>

--
Yann Chemin
Senior Spatial Hydrologist
www.csu.edu.au/research/icwater
M +61-4-3740 7019

7

Yann Chemin wrote:

Thank you Glynn that seems to be resolved,
A last error comes up now:

------------------------

root@SeeePC:/home/yann/grass_yann/imagery/i.landsat.toar# make
VERSION_NUMBER=7.0.svn
/home/yann/grass_yann/dist.i686-pc-linux-gnu/tools/g.html2man.py
/home/yann/grass_yann/dist.i686-pc-linux-gnu/docs/html/i.landsat.toar.html
/home/yann/grass_yann/dist.i686-pc-linux-gnu/man/man1/i.landsat.toar.1
/home/yann/grass_yann/dist.i686-pc-linux-gnu/docs/html/i.landsat.toar.html:133:0:
Error (IndexError('pop from empty list',)): </p>

g.html2man.py requires that the HTML file is structurally valid.

The i.landsat.toar.html file isn't valid; it contains bogus </p> tags.
This appears to arise from the use of <ul>, which is a block element
and thus cannot occur within <p>. The effecti is as if a </p> tag was
inserted immediately before the <ul> tag. The subsequent explicit </p>
tag doesn't match any open <p> tag, resulting in the error.

You can fix the error by adding a <p> tag immediately after the </ul>.

For more information, see the bottom of:

http://grass.osgeo.org/wiki/Updating_GRASS_Documentation

--
Glynn Clements <glynn@gclements.plus.com>

8

Glynn:

g.html2man.py requires that the HTML file is structurally
valid.

The i.landsat.toar.html file isn't valid; it contains bogus
</p> tags.

should be fixed in r43933

Hamish