> On Feb. 8, 2011, 9:48 a.m., Alain Linden wrote:
> > Should AUTOBUILD_CONFIG_FILE be renamed to AUTOBUILD_CONFIG_FILENAME to 
> > clarify this is a name and not a full file path?  Not sure what would 
> > happen if someone set 
> > AUTOBUILD_CONFIG_FILE=foo/bar/baz/MyAutobuildConfig.xml, but probably not 
> > what the user expects.

The option name was chosen to match the name of the long form option that it 
supplies a default value for.

I just tested what would happen, and it's identical to what happens if you use 
the command
  autobuild configure --config-file foo/bar/baz/MyAutobuildConfig.xml
which is to say that the autobuild command reads the configuration file just 
fine, but cmake fails because the relative paths are incorrect.

If this is a problem, I think that it's a separate issue; the environment 
variable is bug-for-bug compatible with the command line switch.


> On Feb. 8, 2011, 9:48 a.m., Alain Linden wrote:
> > autobuild/configfile.py, line 44
> > <http://codereview.secondlife.com/r/140/diff/1/?file=802#file802line44>
> >
> >     This gets called once on package load.  I guess that's fine as long as 
> > you just use autobuild through the autobuild script, but you might get 
> > unexpected results if you are calling autobuild functions in your own 
> > interactive python session. Admittedly that's a corner case...

According to its documentation, the os.environ.get is always going to return 
the same value that it did on the first call regardless.


> On Feb. 8, 2011, 9:48 a.m., Alain Linden wrote:
> > autobuild/configfile.py, line 214
> > <http://codereview.secondlife.com/r/140/diff/1/?file=802#file802line214>
> >
> >     This isn't a warning.  It is perfectly OK to run autobuild with no 
> > script file, like, for example when you call autobuild edit to begin 
> > configuring an autobuild package.

That depends on what you think that 'warning' means... in this case, I think it 
means that if I specified a file that I think exists but it does not, then 
without this warning that fact is not at all obvious - I get an error about 
some package not being defined, not a message that gives me any clue that the 
configuration file is not read.  

I don't think that anyone who is trying to create a new file would be bothered 
by a message that says that it does not already exist at the beginning, but in 
fact I tested it and that's a different code path - you don't get the warning 
when creating the file.


- Oz


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
http://codereview.secondlife.com/r/140/#review343
-----------------------------------------------------------


On Feb. 6, 2011, 5:39 p.m., Oz Linden wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> http://codereview.secondlife.com/r/140/
> -----------------------------------------------------------
> 
> (Updated Feb. 6, 2011, 5:39 p.m.)
> 
> 
> Review request for Viewer.
> 
> 
> Summary
> -------
> 
> This change allows the environment variable AUTOBUILD_CONFIG_FILE to specify 
> a default config-file, so that the precedence becomes:
> 
>    1. the --config-file command line option
>    2. the environment variable AUTOBUILD_CONFIG_FILE
>    3. "autobuild.xml"
> 
> It also adds an info-level (--verbose) display of the config file name that 
> is being loaded, and a warning level display if the config file does not 
> exist.
> 
> 
> This addresses bug open-7.
> 
> 
> Diffs
> -----
> 
>   autobuild/autobuild_tool_build.py 9ee2db08d677 
>   autobuild/autobuild_tool_configure.py 9ee2db08d677 
>   autobuild/autobuild_tool_edit.py 9ee2db08d677 
>   autobuild/autobuild_tool_install.py 9ee2db08d677 
>   autobuild/autobuild_tool_installables.py 9ee2db08d677 
>   autobuild/autobuild_tool_manifest.py 9ee2db08d677 
>   autobuild/autobuild_tool_package.py 9ee2db08d677 
>   autobuild/autobuild_tool_print.py 9ee2db08d677 
>   autobuild/autobuild_tool_uninstall.py 9ee2db08d677 
>   autobuild/configfile.py 9ee2db08d677 
> 
> Diff: http://codereview.secondlife.com/r/140/diff
> 
> 
> Testing
> -------
> 
> Manually tested with and without the command line option, and with and 
> without the environment variable set: confirmed that the correct file name is 
> used per the precedence above.
> 
> Confirmed that the new logging works correctly.
> 
> 
> Thanks,
> 
> Oz
> 
>

_______________________________________________
Policies and (un)subscribe information available here:
http://wiki.secondlife.com/wiki/OpenSource-Dev
Please read the policies before posting to keep unmoderated posting privileges

Reply via email to