On 26/07/13 02:15, Hua Xiang wrote:
Where I can find the standard procedure to create my plugin? Or, should I
create these directory structure manually?
I'm afraid it is a bit of a manual process.. I tend to just create
everything required then svn add everything later but if the directories
don't already exist you can also do something like
svn mkdir bloodhound_time_series_reports/bhtsreports --parents
svn commit bloodhound_time_series_reports -m "adding basic directory
structure for bep-0008 plugin"
to get the bare minimum directories into your branch - you are free to
change either of the "bloodhound_time_series_reports" and "bhtsreports"
directory names but I will keep referring to the directories with that
name in this email.
Anyway, you will be putting a setup.py file in
/home/xiang/bep_0008_time_series_reports/bloodhound_time_series_reports
so that it will be possible to install the plugin but real code for your
project should go in
/home/xiang/bep_0008_time_series_reports/bloodhound_time_series_reports/bhtsreports.
The directories and the updated setup.py file have been committed.
Looks like you have only committed the directories, possibly because I
didn't mention that you also need to "svn add" files so that subversion
knows which files you intend to add to the repository. You can svn add
recursively though sometimes this will add more things than you
intended. An example of this happening in our repository is (at the time
of writing anyway - revision 1507215 has this but I hope that Antonia
will fix it soon) demonstrated in the
bep_0007_embeddable_objects/bloodhound_embedding_plugin directory where
there are at three directories that don't belong in the repository
(BloodhoundEmbeddingPlugin.egg-info, build and dist).
Anyway, all I am saying is that you can either svn add everything and
then use svn stat to try to work out if there is anything you added
accidentally or do the complete opposite and use svn stat to help you
pick out particular files you need to svn add.
I hope that is clear enough for now!
You can use
/home/xiang/bep_0008_time_series_reports/bloodhound_relations or any of
the other bloodhound plugins to give example setup.py files but let us
know if you have any more problems.
I have checked the files in bloodhound_relations directory. I would move
out the changes I committed to the trunk. Just want to know how can I bulid
my plugin and check the display in browser?
Cheers!
A properly written setup.py file will allow you to do this in your
activated virtualenv:
python setup.py develop
so that is useful to do while you are trying to get the setup.py right.
Once it does work, adding the following line to your the
installer/requirements-dev.txt file in your branch:
-e ../bloodhound_time_series_reports
will mean that you and others can install it in the same develop mode
with a "pip install -r requirements-dev.txt".
The question of how to get a report into the system is an interesting
one. As you may be able to tell, generally these are provided at install
time. You could try to implement the IEnvironmentSetupParticipant
interface to give you the opportunity to install your report. You will
not be able to rely on the id for the report being the same on all
installations so you might want to maintain your own database version to
track whether you need to request the upgrade though there may be other
ways of unambiguously determining whether a particular report is installed.
Cheers,
Gary
