wiki:TracJsGanttPlugin

Trac jsGantt plugin

Description

A plugin which allows Trac ticket data to be displayed in a jsGantt chart in a wiki page. Tasks and milestones are links to the corresponding ticket or milestone.

Configurable field names allow integration with other plugins such as MasterTicketsPlugin (for dependencies), SubticketsPlugin (for parent/child relationships) and TimingAndEstimationPlugin (for estimated and total hours).

Over time, the project management aspects of the Gantt chart have been isolated in the TracPM module which has grown to include a query helper (which allows finding tickets based on dependencies) and a ticket change listener (which reschedules tickets based on various constraints) and other support functions. Ultimately, these features -- which really have nothing to do with the Gantt chart -- will be moved to a separate TracPM plugin.

Usage

[[TracJSGanttChart(sample=1)]] displays the sample project from jsgantt.com. [[TracJSGanttChart(milestone=Test)]] displays all the tickets in the Test milestone.

jsGantt sample tasks in a wiki page

[[TracJSGanttChart(goal=1234,schedule=asap)]] finds all the tickets required for ticket 1234 (by following the Required For dependency) and schedules them As Soon As Possible. (In a ticket description or comment, the ticket ID may be replaced by self.)

[[TracJSGanttChart(root=self)]] (when used in a ticket description or comment) finds all the tickets which are descendants of the current ticket (by following the parent/child relationship) and scheduled them As Late As Possible (the default algorithm). (self can be replaced by one or more pipe-delimited ticket IDs as for the id option, e.g., root=1|3|5|9.)

[[TracJSGanttChart(scheduled=1,display=owner:chrisn)]] find all the scheduled tickets and display only those owned by user chrisn.

Arguments

The chart display can be controlled with a number of macro arguments:

Argument Version Description Default
caption Caption to place to right of tasks: None, Caption, Resource, Duration, Complete Resource
comp Show (1) percent complete column, or do not (0). 1
colorBy Field to use to color tasks. Useful fields are priority, owner and milestone but any field can be used. When colored by priority colors are consistent with the colors used in Trac reports. Other coloring choices (e.g., by milestone or owner) use arbitrary, unique colors. priority
dateDisplay Format to display dates: mm/dd/yyyy, dd/mm/yyyy, or yyyy-mm-dd mm/dd/yyyy
display0.10Filter for limiting display of tickets. owner:fred shows only tickets owned by fred. status:closed shows only closed tickets.None
doResourceLeveling0.10Resolve resource conflicts (1) or not (0) when scheduling tickets.0
dur Show (1) duration colunn, or do not (0). 1
endDate Show (1) end date column, or do not (0). 1
expandClosedTickets0.9Show (1) children of closed tickets in the task hierarchy or collapse the subtree (0). 1
format Initial display format, one of those listed in formats day
formatsFormats to show for Gantt chart. A pipe-separated list of minute, hour, day, week, month, and quarter. day|week|month|quarter
goal 0.9 Ticket(s) to show predecessors of. When using something like MasterTicketsPlugin to maintain ticket dependencies, you may create a Gantt showing a ticket and all of its predecessors with goal=<ticket#>. The macro uses the configured succ field to find all predecessor tasks and build an id= argument for Trac's native query handler.

Multiple goals may be provided like goal=1|12|32.

When used in a ticket description or comment, goal=self will display the current ticket's predecessors.
None
hoursPerDay Hours worked per day 8.0
lwidth Width (in pixels) of left table (The one contains task names, etc. on the left of the Gantt chart.) None
omitMilestones0.8Show milestones for displayed tickets (0) or only those specified by milestone= (1). 0
openLevel How many levels of task hierarchy to show open. 1 = only top level task. 999
res Show (1) resource column, or do not (0). 1
root Ticket(s) to show descendants of. When using something like Subtickets plugin to maintain a tree of tickets and subtickets, you may create a Gantt showing a ticket and all of its descendants with root=<ticket#>. The macro uses the configured parent field or parentchild relation to find all descendant tasks and build an id= argument for Trac's native query handler.

Multiple roots may be provided like root=1|12|32.

When used in a ticket description or comment, root=self will display the current ticket's descendants.
None
sample Display (1) sample tasks in Gantt, or do not (0) 0
schedule Schedule algorithm: as-last-as-possible (alap) or as-soon-as-possible (asap) alap
showdep Show (1) dependencies in Gantt, or do not (0). 1
startDate Show (1) start date column, or do not (0). 1
userMap0.8Map (1) user IDs to full names, or do not (0). 1
orderOrder of fields used to sort tickets before display. order=milestone sorts by milestone. May include ticket fields, including custom fields, or "wbs" (work breakdown structure).wbs
scrollToDate to scroll to when chart is initially drawn. A date in yyyy-mm-dd format or "today".None
linkStyleHow to handle clicks on task in the chart. JSGantt style opening a dedicated window (jsgantt) or standard ticket link behaviour (standard).jsgantt

Site-wide defaults for macro arguments may be set in the [trac-jsgantt] section of trac.ini. Where option.<opt> overrides the built-in default for <opt> from the table above.

All other macro arguments are treated as TracQuery specification (e.g., milestone=MS1|MS2) to control which tickets are displayed.

Interfaces

The TracPM module provides several interfaces that can be implemented to adapt the Project Management (PM) features to local business rules. (The TracPM module will be a separate plugin in the future.) These interfaces are defined in pmapi.py.

ITaskSorter

TracPM defers decisions about ticket ordering to an ITaskSorter implementation.

  • Before any comparison is done, ITaskSorter.prepareTasks() is called so that complex keys can be precomputed, external data can be prefetched, etc. to make the comparisons faster and easier.

    prepareTasks() is passed a hash of ticket (tasks). The index of the hash is the ticket ID. The elements of the hash are hashes of Trac ticket attributes. They are not Trac ticket objects.

    prepareTasks() can add to or modify attributes of the ticket. The changes or additions will be available through the life of the scheduling process but will be safely removed before the tickets are returned to the caller.
  • When making decisions about which ticket to schedule first, TracPM sorts the candidate tickets with ITaskSorter.compareTasks().

    compareTasks() is passed two hashes, the attributes of the two tickets to compare. Any attributes added or changed by prepareTasks() are available to compareTasks().
class ITaskSorter(Interface):
    # Process task list to precompute keys or otherwise make
    # compareTasks() more efficient.
    def prepareTasks(self, ticketsByID):
        """Called to prepare tasks for sorting."""

    # Provide a compare function for sorting tasks.
    # Maybe be used as cmp argument for sorted(), and list.sort().
    # Returns -1 if t1 < t2, 0 if they are equal, 1 if t1 > t2.
    def compareTasks(self, t1, t2):
        """Called to compare two tasks"""

Provided Implementations

TracPM provides two implementations of ITaskSorter:

  • SimpleSorter
    • prepareTasks() prefetches the numeric priority values and gives an average priority to any ticket without one.
    • compareTasks() compares tickets based only on their numeric priority
  • ProjectSorter
    • prepareTasks() prefetches the numeric priority values and computes an "effective priority" for each ticket that takes into account its parent's ticket's priority. A high priority child of a low priority parent has a lower effective priority than a low priority child of a high priority parent.
    • compareTasks() compares tickets based only on their effective priority.

SimpleSorter is used by default if no sorter is enabled in trac.ini.

Support Functions

SimpleSorter and ProjectSorter are both derived from BaseSorter which provides several functions which may be useful in implementing custom sorters.

  • _buildEnumMap(field) gets numeric values from the Trac database for enums such as priority and severity. The built-in sorters use it to retrieve values for the priority field in their constructor. It returns a hash, containing numeric values and indexed by name (e.g., 'major').
  • averageEnum() computes the average value in the hash returned by _buildEnumMap().
  • compareOneField(field, t1, t2) compares two tickets based on just one field.

IResourceCalendar

TracPM defers decisions about resource availability to an IResourceCalendar implementation.

  • When scheduling a task, TracPM calls IResourceCalendar.hoursAvailable() to determine how much of a task can be done on a date.
class IResourceCalendar(Interface):
    # Return the number of hours available for the resource on the
    # specified date.
    def hoursAvailable(self, date, resource = None):
        """Called to see how many hours are available on date"""

Provided Implementations

TracPM provides one implementation of IResourceCalendar

  • SimpleCalendar
    • hoursAvailable() returns 8 for Monday through Friday and 0 for Saturday and Sunday.

TeamCalendarPlugin provides an implementation which returns 0 hours for any day that a developer is marked unavailable in the calendar.

ITaskScheduler

An interface for task schedulers exists but it not yet well documented.

class ITaskScheduler(Interface):
    # Schedule each the ticket in tickets with consideration for
    # dependencies, estimated work, hours per day, etc.
    #
    # Assumes tickets is a list returned by TracPM.query().
    #
    # On exit, each ticket has a start and finish that can be accessed
    # with TracPM.start() and finish().  No other changes are made.
    def scheduleTasks(self, options, tickets):
        """Called to schedule tasks"""

Scheduling

If the TicketRescheduler module is enabled, TracPM maintains private tables with task scheduling information. The schedule table has the current schedule:

 Column |  Type   | Modifiers 
--------+---------+-----------
 ticket | integer | not null
 start  | bigint  | 
 finish | bigint  | 
Indexes:
    "schedule_pkey" PRIMARY KEY, btree (ticket)
    "schedule_ticket_idx" btree (ticket)

The schedule_change table holds a record of changes in the schedule over time.

  Column   |  Type   | Modifiers 
-----------+---------+-----------
 ticket    | integer | not null
 time      | bigint  | not null
 oldstart  | bigint  | 
 oldfinish | bigint  | 
 newstart  | bigint  | 
 newfinish | bigint  | 
Indexes:
    "schedule_change_pk" PRIMARY KEY, btree (ticket, "time")
    "schedule_change_ticket_idx" btree (ticket)
    "schedule_change_time_idx" btree ("time")

It is possible -- but not yet supported -- to use the data in schedule_change to reconstruct a baseline schedule for an effective date and then report changes from that baseline.

These tables are maintained by a ticket change listener which looks for changes to fields which might affect schedule then:

  • identifies all tickets that were active (required for an open milestone or an active ticket of the configured milestone ticket type)
  • identifies all tickets that are now active (as a result of the changes)
  • removes any tickets that were active but are no longer from schedule (updating the history in schedule_change)
  • runs the scheduling algorithm on all tickets that are now active
  • writes any changes to schedule (updating history in schedule_change)

With this data in place, you can show a Gantt of scheduled tickets with dates as in the database ([[TracJSGanttChart(scheduled=1)]]) or create reports of the scheduled tickets.

SELECT 
   p.value AS __color__, 
   t.id AS ticket, 
   summary, 
   priority, 
   description as _description,
   est.value as est, 
   act.value as act, 
   -- This bit is specific to Postgres.
   (SELECT to_date(to_timestamp((sched.finish/1000000))::text,
                                'YYYY-MM-DD')) AS due,
   status
FROM ticket AS t
INNER JOIN schedule AS sched ON (t.id=sched.ticket)
INNER JOIN enum AS p ON (p.name=t.priority AND p.type = 'priority')
LEFT OUTER JOIN ticket_custom AS est ON
     (t.id=est.ticket AND est.name='estimatedhours')
LEFT OUTER JOIN ticket_custom AS act ON
     (t.id=act.ticket AND act.name='totalhours')
WHERE t.owner = '$USER' AND status <> 'closed' 
ORDER BY due NULLS FIRST, p.value

Installation

  1. Install globally with:
    sudo easy_install https://trac-hacks.org/svn/tracjsganttplugin/0.11/
    
  2. Enable the plugin by updating TracIni file (..../trac.ini) as follows:
    [components]
    tracjsgantt.* = enabled
    
  3. Configure the project management support for the plugin in its own configuration section, placed into 'trac.ini' file as follows:
    [TracPM]
    # To work with TimingAndEstimationPlugin for percent complete, define ticket fields to use as the data source for:
    fields.estimate = estimatedhours
    fields.worked = totalhours
    
    # Each unit in estimate is 1/8 of a day days_per_estimate = 0.125
    
    # To work with MasterTicketsPlugin for dependencies, define ticket fields to use as the data source for predecessor (pred) and successor (succ).
    fields.pred = blockedby
    fields.succ = blocking
    
    # Alternatively, configure a pred-succ relation like:
    #    relation.pred-succ = mastertickets,source,dest
    # This causes TracPM to query the mastertickets table rather than
    # parse blockedby and blocking custom fields.  This can be somewhat
    # faster.
    
    # To work with SubticketsPlugin for parent/child relationships, 
    # Ticket field to use as the data source for the parent 
    fields.parent = parents
    
    # Alternatively, configure a parent-child relation like:
    #    relation.parent-child = subtickets,parent,child
    # This causes TracPM to query the subtickets table rather than parse 
    # the `parents` custom field.  This can be somewhat faster.
    # When using SubticketsPlugin via a parent-child relation, do not 
    # configure a parent_format (next).
    
    # To work with ChildTickets plugin parent_format, '#%s',
    # Format of ticket IDs in parent field (default: %s).
    parent_format = %s
     
    # Custom fields for start and due dates
    # Ticket field to use as the data source for start date (default: None)
    fields.start = userstart
    # Ticket field to use as the data source for finish date (default: None)
    fields.finish = userfinish
    
    # Format for ''start'' and ''finish'' date strings (default: '%Y-%m-%d')
    date_format = %Y-%m-%d
    
    # Ticket type for milestone-like tickets (default: 'milestone')
    # Used to be milestone_type, that setting is now deprecated.
    goal_ticket_type = milestone
      
    # Ticket field to use as the data source for the percent complete column (default: None).
    fields.percent = complete
       
    # Hours represented by each unit of estimated work (default: 1).
    hours_per_estimate = 1
       
    # Default work for an unestimated task, same units as estimate  (default: 4.0).
    default_estimate = 4.0
       
    # How much work may be remaining when a task goes over estimate, same units as estimate (default: 0.0)..
    estimate_pad = 0.0
    
    • See configuration details below for explanations and more options.
    • Additionally, site-wide defaults for macro arguments may be set at [trac-jsgantt] section. More details about them see #Arguments section.
      [trac-jsgantt]
      option.formats = day|week|month|quarter
      option.format = month
      ## How and which 'columns' to show
      option.lwidth = 300
      option.res = 0
      option.dur = 0
      option.comp = 0
      option.startDate = 0
      option.endDate = 1
      option.dateDisplay = yyyy-mm-dd
      ## How and what to show on Gantt graph
      option.showdep = 1
      option.expandClosedTickets = 1
      option.schedule = asap
      option.openLevel = 0
      option.colorBy = priority
      option.userMap = 0
      option.omitMilestones = 0
      option.caption = Resource
      option.hoursPerDay = 8.0
      
  4. Restart web server on command line:
    $ sudo /etc/init.d/apache2 restart
    

Configuration details

TracJsGanttPlugin is intended to be flexible enough to get data from various plugins by configuring the field names for those plugins in trac.ini. It is known to work with TimingAndEstimationPlugin (for estimated and total hours), MasterTicketsPlugin (for FS dependencies), and SubticketsPlugin for parent/child relationships. Custom fields for start and finish date are also supported.

All of the fields.* items name custom fields which may contain data for the Gantt.

  • When fields.estimate and fields.worked are both configured, the plugin attempts to display (100 * fields.worked/fields.estimate) as the percent complete. The example works with TimingAndEstimationPlugin. Alternatively, if percent is configured, the plugin attempts to display it as the percent complete (it should be a number from 0 to 100). If none of those are configured, all tasks will be marked as 0% complete.
  • When fields.pred and fields.succ are configured the plugin uses them to determine the task dependencies. The example works with MasterTicketsPlugin. If these fields are not configured, no dependencies are shown.
  • When fields.parent is configured, it is the field which holds the parent ticket number. The example works with SubticketsPlugin. If this field is not configured, no parent/child relationship will be displayed. If it is configured, the Gantt can be collapsed by the user to show or hide subtasks. (parent_format determines the format of the content of the parent field. Use "%s" (default) for SubticketsPlugin, or "#%s" for ChildTicketsPlugin.)
  • When fields.start and fields.finish are configured, the plugin uses them to set task start and finish dates. The date_format field is a Python strptime() format specifier which describes the contents of fields.start and fields.finish. If these fields are not configured, all tasks end today and have a 1-day duration.
  • When fields.estimate and fields.finish are both configured (and fields.start is not configured or not on the ticket), the plugin attempts to determine the start of the task from fields.finish and fields.estimate as start = fields.finish - fields.estimate with consideration for weekends and hours per day.
  • The goal_ticket_type may be used to have a custom ticket type show up as milestones on the chart. If this field is not specified, only Trac milestones are displayed as milestones.

Bugs/Feature Requests

Existing bugs and feature requests for TracJsGanttPlugin are here.

If you have any issues, create a new ticket.

Known issues

  • None

Planned enhancements

  1. Display a legend of task colors and their meaning (e.g., which milestone or owner they represent).
  2. Allow some tasks to be open or closed by default.
  3. Display critical path
  4. Display slack time

Source & Download

You can check out TracJsGanttPlugin from here using Subversion, or browse the source with Trac.

Download the zipped source from here.

Recent Changes

[14075] by ChrisNelson on 2014-08-07 14:39:09
Save schedule under Trac 1.0, bump version. Refs #11773.

NOTE: This is incompatible with versions of Trac prior to 1.0!

Plugin version now 0.11. That's a little confusing since plugin 0.10
worked in Trac 0.11(.6). Sorry. This is temporary. Hopefully I'll get
some other changes in soon and bump the plugin version again.

There are other database accesses that should be converted to 1.0 style
but I haven't fully tested those changes yet.

[14016] by rjollos on 2014-07-23 21:28:33
Added license text and file headers. Tidied up indentation. Refs #11871.
[13868] by ChrisNelson on 2014-04-24 16:38:38
Use executemany() when inserting multiple rows. Refs #11027.

Required for cross-db compatibility.

Light testing shows this still works for me in PostgreSQL so it
doesn't seem I broke anything. I need SQLite feedback, though.

[13853] by ChrisNelson on 2014-04-16 22:43:53
Remove comment that has been addressed. Refs #11489.
[13852] by ChrisNelson on 2014-04-16 22:41:44
Handle collapsing groups on multi-Gantt pages. Refs #11489.

Needed to add the chart ID to more DOM element IDs for uniqueness.

[13842] by ChrisNelson on 2014-04-09 13:55:57
Don't schedule closed tickets in background rescheduler. Refs #9648.

Also add some logging.

[13841] by ChrisNelson on 2014-04-09 13:55:54
Simplify interface to prune and repair routines. Refs #9648.
[13840] by ChrisNelson on 2014-04-09 13:55:51
Factor our graph repair. Refs #9648.
[13839] by ChrisNelson on 2014-04-09 13:55:48
Rewrite algorithm to identify tickets to reschedule. Refs #9648.

This is a shorter, cleaner algorithm:

  • Find what's active now
  • Find what was active
  • wasActive - isActive should be idled
  • isActive should be rescheduled
    • Save only if schedule changes

It also works. ;-) With the old algorithm if you has two goals scheduled
and interrupted the earlier one, the tasks for the second goal weren't
moved up. The code to find active and idled tickets was weak and
fragile and not worth fixing since it was also slow. In general, this
new approach handles a lot less data, defers getting ticket details as
long as possible, and gets them for many fewer tickets. In my
playground, I used to query 150 tickets (basically, everything in the
database) and now I query 4-5 tickets (the active set). I expect that
ratio to scale to production. Since querying for ticket details was the
longest part of rescheduling, this is much, much faster than the old
approach.

[13838] by ChrisNelson on 2014-04-09 13:55:45
Factor out schedule and schedule_change update. Refs #9648.

This makes the code which finds affected tickets clearer, closer.

Author/Contributors

Author: ChrisNelson
Maintainer: ChrisNelson
Contributors: rjollos, bof, Matt Sable

Last modified 3 months ago Last modified on Jun 6, 2014 4:16:47 PM

Attachments (1)

Download all attachments as: .zip