Rough notes from Wednesday's Developer Meeting sessions on Wednesday, 2012-02-08
| Table of Contents |
|---|
Implementations, methods, tools
...
(
...
13:15-14:30)
MMI
Working on deployment from 1.5-on
Have two servers
Someone else doing imports in different time zone
...
a lot of Jesse's notes are on the deployer wiki, some are now outdated
useful stuff like "how do I make a different field show up on the search results"
- How do we make it easier to take advantage of each others' notes / documentation
Currently using CollectiveAccess, kind of tailored to our needs
...
- Multiple language
- Guide user behavior
Automatic start/end date generation based on more simple input; e.g. 20th century, generate earliest and last dates
Quarter, half, middle, end, decade, period, century, based on requests from staff at the museum, more or less copied from existing system
Boundary checking for dates
Requires:
- Extensions
- Date library for date arithmetic
- Use case for replacing widget or customizing widget
Integrations we're waiting to do
AD
DAM
Exhibition management
Public web browsing
Export to Regin
Condition database
Literature references - downloads of references to items in the collection from the National Library, pulling them in automatically
ConservationSpace
...
R1 universities with teams that support museums and research collections
Development process, tools, repos (14:30-15:15)
Two areas with a bulk of app layer tests
cspi-services
most important are xml/json two-way conversions
selection of hard-coded service and UI payloads
tomcat-main
focus more on working on the app layer stack, going out to UI/app
start with JSON, will it go out and back with a 200, 400 or whatever
app layer is more of a responsive development mentality
"this needs to happen," and I try to make it happen
try to keep things simple, when we see a pattern, abstract it
cspi-webui
need to rationalize uispec and uischema creation
more standard now, can refactor based on that
will become easier to see how things are done, extend things
hope is that no matter how complex a widget is,
you should be able to do this via configuration
weirdness in structured dates, blobs and medias that needs unifying
app can import an authority via the app layer from a text file
we need to document this and other 'magic' commands better
eGit plug-in for Eclipse http://www.eclipse.org/egit/
Yura mostly using command line for more complicated stuff, e.g. combining branches or contributions
Good UI client
Yura mentioned Atlassian tool used by Toronto, SourceTree http://www.atlassian.com/software/sourcetree/overview
Per Chris/Nate/Yura, GitHub's own tool, GitX, is somewhat disappointing
Patrick has had some issues with Tortoise for git, feels not as good as Tortoise for SVN
gitFlow, abstraction with making merging and branches a little nicer
git start branch
git finish branch
workflow shortcuts, makes what is hard to remember easier https://github.com/nvie/gitflow
Introduction: http://jeffkreeftmeijer.com/2010/why-arent-you-using-git-flow/
Strategies for extensions (15:25-17:00)
The following notes for this session are sparse - randomly selective
What support GitHub provides for community contributions
tagging or naming conventions
Nate: Would likely coalesce around a "good enough" schema, use it thereafter
(Jesse mentioned that phrase)
Patrick:
Business-to-business transfer of loans
Possibility of pluggable export tools, like reports and batch processes
Templates might consist only of language translations of message bundles
Templates for ETL tools - at least as an example; even a bunch of Perl scripts
Local development in the UI layer with record.html
Missing from the mini-build
Ubuntu
has done a good job of distinguishing between:
- this is core
- this is community
- this is the wild west
Django
everything's in the contrib directory
when you check it out, you get all that
A big question / assumption is that you can get your data back out, not just a database dump
Hosting providers can potentially do a better job of backing up, resting restores, etc.
Generating from app layer config might help simplify community contributions