stevenyvr987 / evergreen-mobile-opac Goto Github PK

The project uses a darcs repository, but it is incrementally mirrored to a git 
repository for pushing onto Google Code. Some tasks for finalizing the work 
flow.

- There is a hint that not all darcs records is mirrored to git.
- How to pull from git and push to darcs?

What steps will reproduce the problem?
1. Click 'Log in' button
2. Fill in an incorrect password
3. Click the 'Log in' button

What is the expected output? What do you see instead?
An error dialogue box is shown with a 'Continue' button. If the browser window 
is narrower than the error box, the box is cut off. On an Android 2.2 device, 
the dialogue does not show the 'Continue' button.

Currently, the settings js file is downloaded using a script tag which makes it 
block the browser from downloading other js files at the same time. Also, as a 
module, settings would not need to be added as properties to the window object 
as they now do.

The fmall.js file is currently loaded to provide information for mapping field 
values to data objects. It now seems to be deprecated in favour of a new HTTP 
service that uses the location /IDL2js.

The project uses Unix makefile to implement build targets and tasks. The make 
command is not readily available in windows workstation, thus making it harder 
to development. Unix make is needed when there is a lot of dependencies to 
manage, but we have recently simplified the make rules by removing a lot of 
dependency tracking.

We could replace make with cake, which comes with the coffee script compiler, 
which runs in a nodejs environment. Nodejs is more available in a windows 
workstation.

An open source license is missing from source code.

Currently, opac software is using the first generation Ajax API, eg, 
catalogue.bclibraries.ca/osrf-gateway-v1. We should convert to the second 
generation API, eg, catalogue.bclibraries.ca/osrf-http-translator.

What steps will reproduce the problem?
1. Git pull the repository
2. Issue the make command to build the default targets

What is the expected output? What do you see instead?
A build for deployment should be created; the high-level document should be 
created. Both targets fail, because the rules are accessing tools or files that 
are symlinked.

The design document includes a section about development tools. Recently, a dev 
folder was created in the repository to include some of the development tools 
as part of software distribution. The design document should be updated with 
this detail.

Currently, using 1.0.1

Issuing 'make docs' should reformat comments in source code to source documents 
in the form of html files in the docs folder, but blank files are built instead.

The design doc is written in asciidoc format and would need to be changed to 
the GC wiki syntax. The upside is that it would be easily viewable; the 
downside is that it would need to be synchronized with the master copy, which 
is part of the source distribution.

Some workflows require the patron to be able to place a copy-level hold on an 
item.

1. The makefile is a standard design to calculate the dependency tree of source 
files to make. However, since we are using the coffee-script compiler, it has 
the ability to determine dependencies on its own. Therefore, the makefile could 
be simplified.

2. Another modification to the makefile is to not make the minified version for 
every compilation of a coffeescript file to javascript, which is using Google's 
Closure compiler written in Java and therefore takes a bit of time to execute. 
We could build the minified version of all javascript files as part of the 
build target.

What steps will reproduce the problem?
1. Perform a catalogue search to show a result list.
2. Click the action button on RHS, which should initiate a title search.

What is the expected output? What do you see instead?
We expect to see a new result list of titles by author. Instead, no action is 
performed.

The error seems to have appeared in 0.9.1.

Currently, it isn't clear when the build was last done on software that is 
hosted on test or production servers, other than a manually inserted version 
number, but this does not change between frequent builds that are pushed to 
servers for testing.

Implement a version of the bookbags feature which is suitable for mobile 
devices.

The challenge is to provide a mobile version of the same feature in the full 
opac. The feature can require a lot of screen area to properly represent the 
information.

What steps will reproduce the problem?
1. Using Google Chrome, open the sitka mobile opac using ?ol=bkct
2. Click search bar
3. Click 'Filter search by' selector

What is the expected output? What do you see instead?
We expect to see the list ordered by hierarchy and alphabetically; instead we 
see '_ _ _ TNRD inter-library Loan' listed first. The error does not occur on 
Firefox.

What steps will reproduce the problem?
1. Maximize size of browser window in a workstation
2. Search the public catalogue
3. Click a cover art thumbnail to show its larger version

What is the expected output? What do you see instead?
The large version zooms to fill the browser window, which is alright if we are 
using a mobile device, but it appears to be too large for a normal workstation.

build/
docs/
doc/*.html
app/js/

What steps will reproduce the problem?
1. Perform a search for a title without cover art.
2. Inspect the container for the title using Firebug or similar tool.

What is the expected output? What do you see instead?
We see an empty image element and the resulting layout is positioned as if 
there is an image, thereby wasting space for list items that don't have 
thumbnails. It would be better to not add an empty image element.

What steps will reproduce the problem?
1. Search the public catalogue
2. Choose one of the result titles
3. Click the down link to go to the next title in the list

What is the expected output? What do you see instead?
Browser title should change to show current count, but does not.

The current mechanism is home-built and would be less maintainable. The only 
difficulty would be in removing three features of the current mechanism.

1. Both the subscriber and the publisher are known to the methods, and the 
subscriber is recorded.

2. If the publisher and subscriber are the same entity, no publication is done.

3. If the callback does not return false, publish will refresh the subscriber 
plugin.

By this, we mean book reviews, publisher's bibliographies, and so on.

The current version when pointing to jQM 1.0 B2 shows a blank screen.

What steps will reproduce the problem?

1. Search for titles that have a lot of e-resources, eg, title search of 
'quantum theory' on catalogue.bclibraries.ca

What is the expected output? What do you see instead?

The result lists many titles with no holdings and the summary lines do not 
indicate them as e-resources. Summary lines and title details should show 
information from the 856 tag in the MARC record.

When preparing to show title details, the HTML MARC record is gotten from the 
server and then parsed. We might as show offer to show the user the MARC record 
as an alternative view.

- In search result list, cover art is not aligned within its div element.

- With user logged in, my accounts summary lines are no longer grouped with the 
search line.

Download Google Analytics javascript (ga.js) and include GA preferences in 
settings.js file. 

The current mobile opac lacks the ability to view or edit a patron profile.

Otherwise known as 'shelf browser'.

What steps will reproduce the problem?
1. Search public catalogue
2. Click a title on result list

What is the expected output? What do you see instead?
Network tab of browser inspector tool shows three HTTP GET requests of the form 
'Request URL:data:,/ _ / X'.

Consider using Qunit and Pavlov, an add-on for Qunit that implements a 
higher-level test language.

An item in the search result list shows (1) a thumbnail of cover art on the LHS 
that the user can click to show an enlarged version; (2) the list entry that 
the user can click to show title details; (3) an action button on the RHS that 
the user can click to perform more actions on the title.

Currently, an item in the checkout or holds list shows a list entry that the 
user can click to select the entry. It would be better to mimic the above 
behaviour, so that the user can see the cover art and title details of the item 
checked out or on hold.

The main difference between the search result list and the checkout or holds 
list, is that the latter need to have checkboxes to allow the user to do a mass 
selection, whereas the search result list does not need to be mass selected.

The main file eg_api.coffee is very large and contains several potential 
modules that could be packaged in their own files.

What steps will reproduce the problem?
1. Search catalogue for a title that has cover art
2. Click the title summary in the result list

What is the expected output? What do you see instead?
A thumbnail version of the cover art should show on the left hand side of the 
details area. It does not show after upgrading to jQuery Mobile 1.0

What steps will reproduce the problem?
1. Login and open holds list
2. Select a suspended hold item and click activate button

What is the expected output? What do you see instead?
The holds list should refresh to show the suspended item go active, but it does 
not. Not sure when this error crept in.

The Download tabs should be enabled and populated with a list software builds, 
probably versions 0.8.9, 0.8.10, and 0.9.1.

When development started using jQueryMobile, the main repository, which was 
tagged at 0.8.9 and using a wireframe design, was branched of. Later, the 
original repository continued onto 0.8.10. The 0.8.10 development needs to be 
brought back to the jQueryMobile repository.

The current mobile opac allows for a limited edit of already placed holds: 
cancel, suspend, and resume. However there are more attributes of a placed hold 
that could be usefully edited.

What steps will reproduce the problem?
1. Log in and log out
2. Search catalogue to produce a multi-page list of results
3. Click next link

What is the expected output? What do you see instead?
The next page of result list should be shown. Instead, the login dialogue pops 
up.

Requested by kcls.org.

This file implements the title details page. It consists of three custom jQuery 
plugins under the control of one master plugin. The plugins need to be 
disassembled into their own files.

Also, data objects shared by the plugins are being stored using jQuery data 
objects; perhaps this should be rethought.

Currently, different customizations are maintained by using different server 
clones of the master repository. But it might more lightweight and easier to 
maintain by using branches, each customization would be a branch, with the 
branch named after the library site, eg, 'sitka', 'kcls'.

What steps will reproduce the problem?
1. Perform a catalogue search for titles with cover art
2. Click the cover art thumbnail

What is the expected output? What do you see instead?
A dialogue box is shown with an enlarged version of the cover art. On a mobile 
device with a narrow screen, the dialogue box scales to fit, but the image 
within does not scale.

What steps will reproduce the problem?
1. Log in and click the fines summary bar
2. Click one of the checkboxes
3.

What is the expected output? What do you see instead?
The checkbox should become selected, ie, show a check in the box, but it does 
not. The same situation occurs for checkouts and holds list. The error appeared 
after the upgrade to using jQuery Mobile 1.0.

The current tool, jMod, is not contemporary and does not help the project's 
build phase. Requirejs is contemporary and will help build javascript files 
into downloadable units.

The on method supersedes delegate as of 1.7

Currently, the two versions are managed within the same repository using 
different index.html files (eg, index_sitka.html and index_kcls.html) and two 
settings.js files (eg, settings_sitka.js and settings_kcls.js). The differently 
named files require two strange rules in the makefile to be invoked to 
finalized a build. While the current method is initially convenient, it is not 
scalable nor maintainable.

It would be better to choose one (eg, Sitka) as the main repository, and to 
make the other use a clone, whereby the index.html and settings.js files are 
identically named but different in content. This leads to better scaling as new 
sites which will want differences in interactive behaviour can be better 
accommodated. However, merging patches from the main repository to clones will 
be an extra chore to do.