Документ взят из кэша поисковой машины. Адрес оригинального документа : http://www.eso.org/projects/dfs/team/paneltool-68-7-beta2-test-report.txt Дата изменения: Fri May 25 19:38:37 2001 Дата индексирования: Sun Apr 13 22:55:20 2008 Кодировка: Поисковые слова: enceladus

TEST REPORT for PanelTool V68-7-beta2
=====================================

Tests performed from 22/05/2001 to 24/05/2001.
Main goal here was to check whether bugs originated on previous
V68-0-beta1 release were correctly fixed (See corresponding Test Report
from 18 April 2001 on the DFS Group Home Page and Release
Note on the USS website).

Tests were performed on following configuration:
ns4{flowmgr}23: uname -a
SunOS ns4 5.6 Generic_105181-19 sun4m sparc SUNW,SPARCstation-20
ns4{flowmgr}24: whoami
flowmgr

Accessing USUPLINUX, with period 68 observing runs.
Tool running half in debug mode, half in NO debug mode.


Main performed tests
--------------------

* run Queries: with default columns and sorting, with every
possible selected column and sorting key.

* select different Panel, Sub-panel, telescope values.

* select different programme type.

* Print button.

* Cutoff Line button.

* Verify button.

* enter values via the voting area, or directly in the Grade and
Sigma fields, and check resulting grade and sorting in the
obs runs grid.

* modify Rank values directly from the Rank "slide ruler".

* File>Quit menu item


Conclusion
----------
This test session represents an effort of about 3 hours spread over
2 working days.

The previous Critical bug is correctly fixed by this release (item
1325: two decimal digits in assigned time caused a Java exception).

No new critical bug was found here.

All other items addressed by this release are fixed (see hereafter).

The only remaining problem is the one about 'save', 'query' and
'cutoff line' performances which still should be improved: even if
performances are better than with previous beta1 release, it still
takes from 5 to 6 seconds to display the cut-off line, and about
3 seconds to save a graded obs run (in NO debug mode), which is
not so good. Is JDBC responsible for such performances ? is there
a way to use other (C++) libraries to retrieve data from DB ?
This is in fact a more general design issue that can also
be applicable to P2PP and OT.

Anyway, current PanelTool/DB performances can be temporarily
accepted for the first operational release : this beta2 should
then be the official PanelTool V68-7 to be delivered as
expected on 28 May 2001.


Detailed tests results from KH
------------------------------
* [1302] Release number is now consistent between delivery dirctory
and main panel title bar.
==> OK

* [1303] Password to server is not displayed in clear in the debug
messages.
==> OK

* [1306] Button "Dismiss" chenged to "Deselect". Corresponding action
includes deselecting all rows in the grid.
==> OK

* [1307] Period list extended up to 80.
==> OK

* [1308] Config directorywas cleaned up.
==> OK

* [1310] VLTI added to telescope list.
==> OK

* [1315,1318] Reports printed in landscape format by default. Font is also
smaller.
==> OK

* [1319] Sorting from column headers is now disabled.
==> OK

* [1321] Performance of cutoff line computation was improved.
==> OK, but still to improve more.

* [1324] When invalid values are entered a warning is shown, then the
entry cell is refreshed.
==> OK

* [1325] Fixed bug: two decimal digits in assigned time would cause an
exception.
==> OK

* [1327] Cutoff conditions are bow verified before running the query.
==> OK

* [1328] Error messages are now shown in a popup panel.
==> OK

* [No SPR] Startup script redirects printed output to a log file
rather than to /dev/null, to allow post-mortem checks.
Log file is created in the startup directory; pathname includes
panel and subpanel designation, as well as process number.
No log file is produced in debug mode.
==> OK, but question: how/who will clean such log files which may
become very big after a while ?!