evergreen:rel_3_1

Last commit made on 2019-08-22
Get this branch:
git clone -b rel_3_1 https://git.launchpad.net/evergreen

Branch merges

Branch information

Name:
rel_3_1
Repository:
lp:evergreen

Recent commits

1f8440d... by Dan Wells 7 hours ago

Forward-port 3.1.14 upgrade script

Signed-off-by: Dan Wells <email address hidden>

42e758e... by Debbie Luchenbill on 2019-04-24

Docs: add checkin trigger holds and cancel transit

Signed-off-by: dluchenbill <email address hidden>
Signed-off-by: Jane Sandberg <email address hidden>

0dee4a7... by Andrea Neiman 9 hours ago

docs: error correction to 3.1.14 release notes

corrected link for "Fixes display issue with menus at certain screen resolutions" - the link & bug number were incorrect.

Signed-off-by: Andrea Buntz Neiman <email address hidden>
Signed-off-by: Jane Sandberg <email address hidden>

5357f13... by Dan Wells on 2019-08-15

LP#1796945 Match new path_label/alias standard

The old reporter used '::' as a path separator in some labels, while the
new version uses '->' in its place. This would normally be just a
curiosity, except that the hashes used as join aliases are generated
from these path labels. The end result is that while a report can be
cloned, if you add a field anywhere other than the top level, it
references a join alias which does not exist, and the report breaks.

Now, the method by which report templates are upgraded from v4 to v5
does not happen all in one pass, but rather a first pass is made
populate the interface, then the rest is done when the user saves the
clone. Because of this, it actually seems adequate to only fix the
label, then allow the other steps to fill in the other bits upon saving
of the report.

Thank you to J. Boyer for helping track down the culprit.

Signed-off-by: Dan Wells <email address hidden>
Signed-off-by: Jason Boyer <email address hidden>

cc444ec... by Dan Wells on 2019-05-03

LP#1796945 Reporter cloning and creation fixes

This commit addresses a variety of issues with the webstaff reporter
interface, particularly cases of cloning reports created in the XUL
client.

1. The conversion process did not account for manually selected JOIN
operations (aka nullability). These JOINs are now honored by the
conversion code.

2. The conversion process did not account for aggregate filters. These
filters are now converted where present.

3. The previous reporter interface attempted to intelligently apply LEFT
and INNER JOINs by default. The new interface applied INNER joins
exclusively by default, leading in many cases to different results.
This commit reinstates the previous logic. One side effect of this
change is that the IDL tree itself is no longer opinionated about JOIN
type, and the default JOIN is undefined.

4. The nullability selector has been expanded to allow for manual
selection of INNER joins, as they will longer be the default in some
cases.

5. Cloned-converted reports did not retain column order. The order is
now preserved.

6. Some templates created in the older interface could, in some cases,
have aggregate values set as the string "undefined" rather than
actually being undefined. This led to converted templates failing with
"column [xxx] must appear in the GROUP BY clause...", as they were
incorrectly converted as aggregates. The conversion code now accounts
for this latent bug.

Signed-off-by: Dan Wells <email address hidden>
Signed-off-by: Jason Boyer <email address hidden>

6d683f8... by Andrea Neiman on 2019-08-21

Docs: Adding 3.1.14 release notes

Signed-off-by: Andrea Buntz Neiman <email address hidden>
Signed-off-by: Jane Sandberg <email address hidden>

697e241... by James Fournie <email address hidden> on 2019-05-02

LP1751800 - fix fields fields reversing

The pop() here will remove the last item in the array, only to add it
later with a push() a few lines later. Instead use a shift() to
preserve the order.

Signed-off-by: James Fournie <email address hidden>
Signed-off-by: Remington Steed <email address hidden>
Signed-off-by: Dan Wells <email address hidden>

c071d06... by Galen Charlton on 2019-05-22

LP#1785061: move the filter value munging to the template service

This allows the reporter app to stay a bit more focused on
display concerns.

Here's a test plan for the patch series
---------------------------------------
[1] Create a reporter template that has a filter field
    and a filter operator of "In list", "Not in list", "Between",
    or "Not between" and hard-code a value for that filter
    in the template, separating values with a comma.
[2] Try to create a report from that template; note that it
    fails with an error.
[3] Apply the patch and repeat steps 1 and 2. This time, the
    report should succeed.

Note that this fix applies only to hardcoding filter values in the
template; it doesn't change any behavior when attempting to set
a filter value for any of the four operators above at the point of
creating a report.

Signed-off-by: Galen Charlton <email address hidden>

Conflicts:
 Open-ILS/web/js/ui/default/staff/reporter/template/app.js

Signed-off-by: Dan Wells <email address hidden>

657577d... by Remington Steed on 2019-05-02

LP#1785061: Split filter value on comma for "in list" and the like

This commit borrows directly from the XUL reporter code (see
function __default_value_event_handler () in
Open-ILS/web/reports/xul/template-config.js). Basically, when the filter
value is saved, certain cases need special treatment, such as splitting
an "in list" value on commas. This commit includes a helper function
which does the special treatment and saves the filter value. This helper
is called both when the value itself is changed, and when the operator
is changed.

Signed-off-by: Remington Steed <email address hidden>
Signed-off-by: Galen Charlton <email address hidden>

Conflicts:
 Open-ILS/web/js/ui/default/staff/reporter/template/app.js

Signed-off-by: Dan Wells <email address hidden>

573a311... by Galen Charlton on 2019-08-06

LP#1722827: stamp DB update

Signed-off-by: Galen Charlton <email address hidden>