/adei/trunk

To get this branch, use:
bzr branch http://darksoft.org/webbzr/adei/trunk
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
Important
=========
 ! downloads_check.php - doesn't handle correctly failed downloads. We should
 do something about it.
 ! For MSSQL column and table quotes should be changed from " to []. Yes, for
 this we should introduce opening and closing quote :(
 ! Prevent entering custom plot mode if not supported

Cache Engine
============
 - We are currently just skipping databases which are failed to connect to the
 data source (request.pph/GetSources/catch block). In case if we are caching the
 data this prevents administrator from getting information about the problems.
 - If we change use_subsecond flag without dropping the cache, the ADEI will
 report channel inconsistency (not clarifying the issue). Since, we will get
 two extra columns in the caching table. The proper error message should be
 reported in this case.
 - At the moment, We can't really use non [\w\d_] symbols in gids if md5 
 postfixes are switched off. Otherwise, this symbols are stripped from the 
 postfixes (for the sake of table names) and then the check for cached tables
 is failing (it reconstructs gid basing on the table names). Some, other things
 may fail as well.
 For the check, we can try to ignore the inappropriate symbols (only if md5 is
 switched off). So, the stripping should be performed by caching code.
 - In the non-0 caching modes, the last interval is not displayed. Why cache is
 not returning the last point?

VIEWS
=====
 - Support non-fitting timestamps in the scatter plot with resolution 0.

Filters
=======
 - While requesting last record (or potentially several last entries), it is
 possible to get empty response if the last record while be rejected by the
 configured filters. Iteratively increase limit? When fixed and may be 
 universally used, shall we introduce a 'limit' option instead of making 
 complex 'window' parsing?
 - If we may want to filter the rows based on value of certain channel, we
 may get problems with db_mask. It will be applied before the filter and
 only selected channels will pushed through. Using READER_Filter interface
 we may find the required columns, but will should ensure that the data
 required for filter is provided (introduce db_mask_required parameter of 
 filter?)


Virtual
=======
 - Support "__" in group names in SourceTree

Download Manager
===============
 ! Download manager does not respect "setup"
 ! Show details of Downloads Manager button while not work if data is not 
 from default setup.
 ! download_manager.js is updating too often (due to adei.updater.Update()
 calls). This should be somehow fixed. As well it refers dlmanager.Request()
 which uses variable name set in correspndent module. This is completely 
 insane and should be fixed.

ZEUS Reader
===========
 ! Interlock between Full Backup and other calls in ZEUSReader. This is
 critical. If not prevented, the cache may start while configuration is
 partially dropped and not yet reloaded. This may result in rejecting 
 some data vectors because of invalid number of records, etc.

RRD Reader
==========
 ! Axes are not always cleaned when changing to different group, 
 with RRD source 
 - Do not list empty groups in RRDReader

Web Frontend
============
 - Rising memory consumption with Firefox (and I guess other browsers)
 - Passing away pointer to "this" in JS-constructors may be problematic (no 
 errors, but this will be an incorrect "this").
 - Chromium and arora doesn't respect disabled states while setting initial 
 choices of select (scatter plot options). Just force proper selection.
 - Opera/Chrome do not resize source popup properly on axes update. If width
 of axes tab grows, it just gets out of popup borders.
 - The following case: We made a selection and opened one of the popups. The
 forced graph update have canceled selection. However, the selected values
 are still present window configuration dialog.  Should we cancel them? 
 Or should we re-draw selection after update?
 - At the moment on the browser window resize the graphic is not autoupdated,
 we probably should force update.
 
Graph
=====
 - With small graph area (and big left part), the time labels are starting
 to overlap at some point (Andreas)
 - After zooming by mouse, when changing to the secondary window or info tab,
 the "Invalid WINDWO + is specified" returne (Andreas)
 - Prevent complete shadowing (filling) of screen with a single chart in the 
 multiaxis mode. Try to find such axes and fix the problem by stacking them 
 vertically and increasing axis range.
 - Logarithmic axes has problems with drawing 0 values, - where should we
 start?