Distributed issue tracking
klog - A simple distributed bug tracking system.
klog is an intentionally simple system which allows bugs to be stored inside projects, and merged in along with all other changes in exactly the way that a user of a distributed revision control system would expect.
In short a project will have any and all bugs stored beneath the .klog directory. These bugs will be stored in random, but hostname-sepcific, filename such that multiple people merging and commiting will be unlikely to ever see conflicts.
Any project may have bugs in two states:
Each bug will have an associated UID and number. The UID associated with a bug will be unchanged and fixed at the time it is created, but the number is a purely local convience.
Usage of klog is divided into several distinct cases:
These actions all work in a consistent manner, to avoid unpleasant suprises.
To initialise a bug database within your project run:
Once you've initialised your klog database you will need to ensure that you add the .klog database to your revision control system.
To add a new bug, optionally specifying a title for it, please run:
klog add This is my bug title
If no title is specified a bug report will be created with a default title. This will open an editor for you to enter the bug report text.
The editer may be specified with the --editor flag, the EDITOR environmental variable, and will otherwise default to vim.
Once a bug report has been created you should find that it is visible in the output of "klog list" or "klog open". In both cases you'll see output which looks something like this:
N:0001 [closed] testing me N:0002 [closed] This is atest N:0003 [ open] This is my bug title
This listing report shows three things: The number of the bug, the state of the bug ("open" vs. "closed") and the title of the bug.
Each of the operations that is specific to a single bug report will allow you to specify the number of the bug. For example if you wished to update the last bug, to append some text to it, you could run:
klog append 3
Similarly you could close the bug by running:
klog close 3
Note that to close a bug you do not need to give a justification, or add any content. A bug may go from freshly opened to closed with no need for further updates.
Internally each bug is stored in a file, beneath the .klog directory.
Each bug file has a random name which is designed to avoid potential collisions if a repository is shared between many users, upon different systems, as is common with distributed revision controls.
Each bug report will have several fixed fields at the beginning, as this example shows:
Title: This is atest UID: 1269990083.P10668M151020.birthday.my.flat Added: Wed Mar 31 00:01:23 2010 Status: open I like pies, but I have none.
The UID is essentially random, but should be unique, and is the portable sane way to refer to bugs. When running klog list you'll see a number reported next to each bug, but this number is valid only for the local system and may change when new bugs are reported.
In short you may use the displayed "bug number" for carrying out local operations providing you realise that the number associated with a specific bug will change over time. By contrast the UID will never change, so you may always run a command like this:
klog view 1270024997.P15442M277230.birthday.my.flat
The template which is presented to the user when they report a new bug may be replaced. If the file ".klog/new-bug-template" is present the contents of that file will be inserted in new reports, rather than the default message.
If the file .klog/hook exists, and is executable, it will be invoked when new bugs are added, bugs are closed, or comments are updated.
The hook will be invoked with two arguments, the first will be a string defining the action which has caused the invocation, the second will be the name of the bug file. For example you might use this to auto-add new bug reports to the repository with a hook like this:
#!/bin/sh if [ "$1" = "add" ]; then hg add "$2" fi