RT 6.0.3 Documentation

RT::StyleGuide

NAME

RT::StyleGuide - RT Developer Guide

INTRODUCTION

All code and documentation that is submitted to be included in the RT distribution should follow the style in this document. This is not to try to stifle your creativity, but to make life easier for everyone who has to work with your code, and to aid those who are not quite sure how to do something.

The conventions below apply to Perl modules, web programs, and command-line programs, specifically, but also might apply to some degree to any Perl code written for use in RT.

Note that these are all guidelines, not unbreakable rules. If you have a need to break one of the rules herein, however, then it is best to first start a discussion in the RT Developers category on the community forum at https://forum.bestpractical.com.

Note that with much of this document, it is not so much the Right Way as it is Our Way. We need to have conventions in order to make life easier for everyone. So don't gripe, and just follow it, because you didn't get a good grade in "Plays Well With Others" in kindergarten and you want to make up for it now.

We don't always follow this guide. We are making changes throughout our code to be in line with it. But just because we didn't do it yet, that is no excuse. Do it anyway. :-)

This document is subject to change at the whims of the core RT team.

DEVELOPMENT SETUP

Setting up a development environment

A quick and easy way to try out changes in RT is to use SQLite and the inplace layout so you can run code inside your git checkout. Here is a sample configuration you can use. Put this in a file under etc/RT_SiteConfig.d, say etc/RT_SiteConfig.d/dev.pm:

    Set( $rtname, 'rt.example.com');
    Set( $CommentAddress, 'rt-comment@example.com' );
    Set( $CorrespondAddress, 'rt@example.com' );
    Set( $Organization, 'example.com' );

    # Set a domain you can hit locally, for example localhost.
    Set( $WebDomain, 'localhost' );
    Set( $WebPort, '8080' );

    # Allows you to run over http without SSL.
    Set($WebSecureCookies, 0);

    # Lets you see all RT log output in your terminal.
    Set( $LogToSTDERR, 'debug');

    # Send email output to files in var so you can look at them.
    Set( $MailCommand, 'mbox');

    # Don't use the Mason cache, so you can see changes immediately.
    Set( $DevelMode, 1);

    # Useful if you need to see raw SQL, but don't leave on because it logs a
    # ton.
    #Set( $StatementLog, 1);

Then run:

    ./configure.ac --with-my-user-group --enable-layout=inplace \
      --enable-developer --with-db-type=SQLite
    make testdeps
    make fixdeps
    make initdb
    sbin/rt-server

You will now be able to access RT at: http://localhost:8080

If the Mason cache is enabled (i.e, you haven't set $DevelMode), then as you make changes you will likely need to clear the cache:

    rm -rf var/mason_data

You may also need to clear the database as you adjust, test things etc. By default, the SQLite database file is created in var/rt5. Just delete that file to clear the test database, and re-run make initdb to start fresh.

You can also run development checkouts with all the databases RT supports. Change the configure.ac options above to set the database you want to use. That will change the dependencies installed by make fixdeps, so you need to re-run that and install the drivers for the new database.

CODING PRINCIPLES

Perl Version

We code everything to Perl 5.26.3 or higher.

Documentation

All modules will be documented using the POD examples in the module boilerplate. The function, purpose, use of the module will be explained, and each public API will be documented with name, description, inputs, outputs, side effects, etc.

If an array or hash reference is returned, document the size of the array (including what each element is, as appropriate) and name each key in the hash. For complex data structures, map out the structure as appropriate (e.g., name each field returned for each column from a DB call; yes, this means you shouldn't use "SELECT *", which you shouldn't use anyway).

Also document what kind of data returned values are. Is it an integer, a block of HTML, a boolean?

All command-line program options should be documented with a command-line help option showing usage and all options. Each available function, switch, etc. should be documented, along with a statement of function, purpose, use of the program. Do not use the same options as another program, for a different purpose.

All web templates should be documented with a statement of function, purpose, and use in a mason comment block.

Any external documents, and documentation for command-line programs and modules, should be written in POD, where appropriate. From there, they can be translated to many formats with the various pod2* translators. Read the perlpod manpage before writing any POD, because although POD is not difficult, it is not what most people are used to. It is not a regular markup language; it is just a way to make easy documentation for translating to other formats. Read, and understand, the perlpod manpage, and ask us or someone else who knows if you have any questions.

Version

Our distribution versions use tuples, where the first number is the major revision, the second number is the version, and third number is the subversion. Odd-numbered versions are development versions. Examples:

        1.0.0           First release of RT 1
        1.0.1           Second release of RT 1.0
        1.0.10          etc.
        1.1.0           First development release of RT 1.2 (or 2.0)
        2.0.0           First release of RT 2

Versions may end in "rc" or "beta" and a number if they are release candidates:

        2.0.0rc1        First release candidate for real 2.0.0

Comments

All code should be self-documenting as much as possible. Only include necessary comments. Use names like "$ticket_count", so you don't need to do something like:

        # ticket count
        my $tc = 0;

Include any comments that are, or might be, necessary in order for someone else to understand the code. Sometimes a simple one-line comment is good to explain what the purpose of the following code is for. Sometimes each line needs to be commented because of a complex algorithm. Read Kernighan & Pike's Practice of Programming about commenting. Good stuff, Maynard.

Warnings and Strict

All code must compile and run cleanly with "use strict" enabled and the perl "-w" (warnings) option on. If you must do something that -w or strict complains about, there are workarounds, but the chances that you really need to do it that way are remote.

Lexical Variables

Use only lexical variables, except for special global variables ($VERSION, %ENV, @ISA, $!, etc.) or very special circumstances (see %HTML::Mason::Commands::session ). Global variables for regular use are never appropriate. When necessary, "declare" globals with "use vars" or "our()".

A lexical variable is created with my(). A global variable is pre-existing (if it is a special variable), or it pops into existence when it is used. local() is used to tell perl to assign a temporary value to a variable. This should only be used with special variables, like $/, or in special circumstances. If you must assign to any global variable, consider whether or not you should use local().

local() may also be used on elements of arrays and hashes, though there is seldom a need to do it, and you shouldn't.

Pass by Reference

Arrays and hashes should be passed to and from functions by reference only. Note that a list and an array are NOT the same thing. This is perfectly fine:

        return($user, $form, $constants);

An exception might be a temporary array of discrete arguments:

        my @return = ($user, $form);
        push @return, $constants if $flag;
        return @return;

Although, usually, this is better (faster, easier to read, etc.):

        if ($flag) {
                return($user, $form, $constants);
        } else {
                return($user, $form);
        }

We need to talk about Class::ReturnValue here.

Method parameters

If a method takes exactly one mandatory argument, the argument should be passed in a straightforward manner:

        my $self = shift;
        my $id = shift;

In all other cases, the method needs to take named parameters, usually using a %args hash to store them:

        my $self = shift;
        my %args = (
            Name => undef,
            Description => undef,
            @_
        );

You may specify defaults to those named parameters instead of using undef above, as long as it is documented as such.

It is worth noting that the existing RT codebase had not followed this style perfectly; we are trying to fix it without breaking existing APIs.

Tests

Modules should provide test code, with documentation on how to use it. Test::More makes it easy to create tests. Any code you write should have a test suite. Any code you alter should have a test suite. If a patch comes in without tests, there is something wrong.

When altering code, you must run the test harness before submitting a patch or committing code to the repository.

See "TESTING".

STDIN/STDOUT

Always report errors using RT->Logger. It's a Log::Dispatch object. Unlike messages meant for the user, log messages are not to be internationalized.

There are several different levels (RT->Logger methods) of logging:

debug

Used for messages only needed during system debugging.

info

Should be used to describe "system-critical" events which aren't errors. Examples: creating users, deleting users, creating tickets, creating queues, sending email (message id, time, recipients), receiving mail, changing passwords, changing access control, superuser logins)

warn/warning

Used to report issues that need attention, but are not failures. Many systems will run production at the "warn" level, so only use this for important events.

error

Used for RT-generated failures during execution.

crit

Should be used for messages when an action can not be completed due to some error condition beyond our control.

In the web UI and modules, never print directly to STDERR. Do not print directly to STDOUT, unless you need to print directly to the user's console.

In command-line programs, feel free to print to STDERR and STDOUT as needed for direct console communication. But for actual error reporting, use the logging API.

System Calls

Always check return values from system calls, including open(), close(), mkdir(), or anything else that talks directly to the system. Perl built-in system calls return the error in $!; some functions in modules might return an error in $@ or some other way, so read the module's documentation if you don't know. Always do something, even if it is just calling RT->Logger->warning(), when the return value is not what you'd expect.

STYLE

Much of the style section is taken from the perlstyle manpage. We make some changes to it here, but it wouldn't be a bad idea to read that document, too.

Terminology

function vs. sub(routine) vs. method

Just because it is the Perl Way (not necessarily right for all languages, but the documented terminology in the perl documentation), "method" should be used only to refer to a subroutine that are object methods or class methods; that is, these are functions that are used with OOP that always take either an object or a class as the first argument. Regular subroutines, ones that are not object or class methods, are functions. Class methods that create and return an object are optionally called constructors.

Users

"users" are normally users of RT, the ones hitting the site; if using it in any other context, specify. "system users" are user names on the operating system. "database users" are the user names in the database server. None of these needs to be capitalized.

Names

Don't use single-character variables, except as iterator variables.

Don't use two-character variables just to spite us over the above rule.

Constants are in all caps; these are variables whose value will never change during the course of the program.

        $Minimum = 10;          # wrong
        $MAXIMUM = 50;          # right

Other variables are lowercase, with underscores separating the words. The words used should, in general, form a noun (usually singular), unless the variable is a flag used to denote some action that should be taken, in which case they should be verbs (or gerunds, as appropriate) describing that action.

        $thisVar      = 'foo';  # wrong
        $this_var     = 'foo';  # right
        $work_hard    = 1;      # right, verb, boolean flag
        $running_fast = 0;      # right, gerund, boolean flag

Arrays and hashes should be plural nouns, whether as regular arrays and hashes or array and hash references. Do not name references with "ref" or the data type in the name.

        @stories     = (1, 2, 3);      # right
        $comment_ref = [4, 5, 6];      # wrong
        $comments    = [4, 5, 6];      # right
        $comment     = $comments->[0]; # right

Make the name descriptive. Don't use variables like "$sc" when you could call it "$story_count". See "Comments".

There are several variables in RT that are used throughout the code, that you should use in your code. Do not use these variable names for anything other than how they are normally used, and do not use any other variable names in their place. Some of these are:

        $self           # first named argument in object method

Subroutines (except for special cases, like AUTOLOAD and simple accessors) begin with a verb, with words following to complete the action. Accessors don't start with "Get" if they're just the name of the attribute.

Accessors which return an object should end with the suffix Obj.

This section needs clarification for RT.

Words begin with a capital letter. They should as clearly as possible describe the activity to be performed, and the data to be returned.

        Load();         # good
        LoadByName();   # good
        LoadById();     # good

Subroutines beginning with _ are special: they are not to be used outside the current object. This is not to be enforced by the code itself, but by someone very big and very scary.

For large for() loops, do not use $_, but name the variable. Do not use $_ (or assume it) except for when it is absolutely clear what is going on, or when it is required (such as with map() and grep()).

        for (@list) {
            print;                      # OK; everyone knows this one
            print uc;                   # wrong; few people know this
            print uc $_;                # better
        }

Note that the special variable _ should be used when possible. It is a placeholder that can be passed to stat() and the file test operators, that saves perl a trip to re-stat the file. In the example below, using $file over for each file test, instead of _ for subsequent uses, is a performance hit. You should be careful that the last-tested file is what you think it is, though.

        if (-d $file) {         # $file is a directory
            # ...
        } elsif (-l _) {        # $file is a symlink
            # ...
        }

Package names begin with a capital letter in each word, followed by lower case letters (for the most part). Multiple words should be StudlyCapped.

        RT::User                        # good
        RT::Database::MySQL             # proper name
        RT::Display::Provider           # good
        RT::CustomField                 # not so good, but OK

Plugin modules should begin with "RT::Extension::", followed by the name of the plugin.

CODE FORMATTING

The RT codebase is more than 25 years old; as such, there are sections which do not (yet) conform to the guidelines below. Please attempt to follow the guidelines, even if the code surrounding your changes does not yet.

RT includes a .perltidyrc in its top-level which encodes many of the conventions. When in doubt, use perltidy.

Indentation and Blank Space

Each level of indentation should be four spaces; hard tabs are forbidden.

No space before a semicolon that closes a statement.

        foo(@bar) ;     # wrong
        foo(@bar);      # right

Line up corresponding items vertically.

        my $foo   = 1;
        my $bar   = 2;
        my $xyzzy = 3;

        open(FILE, $fh)   or die $!;
        open(FILE2, $fh2) or die $!;

        $rot13 =~ tr[abcedfghijklmnopqrstuvwxyz]
                    [nopqrstuvwxyzabcdefghijklm];

        # note we use a-mn-z instead of a-z,
        # for readability
        $rot13 =~ tr[a-mn-z]
                    [n-za-m];

Put blank lines between groups of code that do different things. Put blank lines after your variable declarations. Put a blank line before a final return() statement. Put a blank line following a block (and before, with the exception of comment lines).

An example:

        # this is my function!
        sub foo {
            my $val = shift;
            my $obj = new Constructor;
            my($var1, $var2);

            $obj->SetFoo($val);
            $var1 = $obj->Foo();

            return($val);
        }

        print 1;

Parentheses

For control structures, there is a space between the keyword and opening parenthesis. For functions, there is not.

        for(@list)      # wrong
        for (@list)     # right

        my ($ref)       # wrong
        my($ref)        # right

Be careful about list vs. scalar context with parentheses!

        my @array = ('a', 'b', 'c');
        my($first_element) = @array;            # a
        my($first_element) = ('a', 'b', 'c');   # a
        my $element_count  = @array;            # 3
        my $last_element   = ('a', 'b', 'c');   # c

Always include parentheses after functions, even if there are no arguments. There are some exceptions, such as list operators (like print) and unary operators (like undef, delete, uc).

There is no space inside the parentheses, unless it is needed for readability.

        for ( map { [ $_, 1 ] } @list ) # OK
        for ( @list )                   # not really OK, not horrible

On multi-line expressions, match up the closing parenthesis with either the opening statement, or the opening parenthesis, whichever works best. Examples:

        @list = qw(
            bar
            baz
        );                      # right

        if ($foo && $bar && $baz
                 && $buz && $xyzzy) {
            print $foo;
        }

Whether or not there is space following a closing parenthesis is dependent on what it is that follows.

        print foo(@bar), baz(@buz) if $xyzzy;

Note also that parentheses around single-statement control expressions, as in if $xyzzy, are optional (and discouraged) if it is absolutely clear -- to a programmer -- what is going on. There is absolutely no need for parentheses around $xyzzy above, so leaving them out enhances readability. Use your best discretion. Better to include them, if there is any question.

The same essentially goes for Perl's built-in functions, when there is nothing confusing about what is going on (for example, there is only one function call in the statement, or the function call is separated by a flow control operator). User-supplied functions must always include parentheses.

        print 1, 2, 3;                          # good
        delete $hash{key} if isAnon($uid);      # good

However, if there is any possible confusion at all, then include the parentheses. Remember the words of Larry Wall in the perlstyle manpage:

        When in doubt, parenthesize.  At the very least it will
        let some poor schmuck bounce on the % key in vi.

        Even if you aren't in doubt, consider the mental welfare
        of the person who has to maintain the code after you, and
        who will probably put parens in the wrong place.

So leave them out when it is absolutely clear to a programmer, but if there is any question, leave them in.

Braces

(This is about control braces, not hash/data structure braces.)

There is always a space before the opening brace.

        while (<$fh>){  # wrong
        while (<$fh>) { # right

A one-line block may be put on one line, and the semicolon may be omitted.

        for (@list) { print }

Otherwise, finish each statement with a semicolon, put the keyword and opening curly on the first line, and the ending curly lined up with the keyword at the end.

        for (@list) {
            print;
            smell();
        }

Generally, we prefer not "cuddled else":

        if ($foo) {
            print;
        }
        else {
            die;
        }

Operators

Put space around most operators. The primary exception is the for aesthetics; e.g., sometimes the space around "**" is omitted, and there is never a space before a ",", but always after.

        print $x , $y;  # wrong
        print $x, $y;   # right

        $x = 2 >> 1;    # good
        $y = 2**2;      # ok

Note that "&&" and "||" have a higher precedence than "and" and "or". Other than that, they are exactly the same. It is best to use the lower precedence version for control, and the higher for testing/returning values. Examples:

        $bool = $flag1 or $flag2;       # WRONG (doesn't work)
        $value = $foo || $bar;          # right
        open(FILE, $file) or die $!;

        $true  = foo($bar) && baz($buz);
        foo($bar) and baz($buz);

Note that "and" is seldom ever used, because the statement above is better written using "if":

        baz($buz) if foo($bar);

Most of the time, the confusion between and/&&, or/|| can be alleviated by using parentheses. If you want to leave off the parentheses then you must use the proper operator. But if you use parentheses -- and normally, you should, if there is any question at all -- then it doesn't matter which you use. Use whichever is most readable and aesthetically pleasing to you at the time, and be consistent within your block of code.

Break long lines AFTER operators, except for ".", "and", "or", "&&", "||". Try to keep the two parts to a binary operator (an operator that has two operands) together when possible.

        print "foo" . "bar" . "baz" .
              "buz";                            # wrong

        print "foo" . "bar" . "baz"
            . "buz";                            # right

        print $foo unless $x == 3 && $y ==
                4 && $z == 5;                   # wrong

        print $foo unless $x == 3 && $y == 4
                       && $z == 5;              # right

Other

Put space around a complex subscript inside the brackets or braces.

        $foo{$bar{baz}{buz}};   # OK
        $foo{ $bar{baz}{buz} }; # better

In general, use single-quotes around literals, and double-quotes when the text needs to be interpolated.

It is OK to omit quotes around names in braces and when using the => operator, but be careful not to use a name that doubles as a function; in that case, quote.

        $what{'time'}{it}{is} = time();

When making compound statements, put the primary action first.

        open(FILE, $fh) or die $!;      # right
        die $! unless open(FILE, $fh);  # wrong

        print "Starting\n" if $verbose; # right
        $verbose && print "Starting\n"; # wrong

Use here-docs instead of repeated print statements.

                print <<EOT;
        This is a whole bunch of text.
        I like it.  I don't need to worry about messing
        with lots of print statements and lining them up.
        EOT

Just remember that unless you put single quotes around your here-doc token (<<'EOT'), the text will be interpolated, so escape any "$" or "@" as needed.

INTERNATIONALIZATION

RT has been translated into several dozen languages. We use Launchpad ( https://translations.launchpad.net/rt ) to crowdsource our translations into po files. RT uses Locale::Maketext to localize its user interface.

Your first stop on this magical journey of internationalization is Locale::Maketext::TPJ13, which explains the whys of Locale::Maketext. RT uses most of the features developed in that article.

Using loc() and /l

Strings that are displayed to users should be passed through the loc("...") function or the <&|/l&>...</&> Mason template. loc and /l both take parameters, which are used in place of string interpolation (much like sprintf). It's acceptable to use HTML in /l calls, especially for bold and emphasis. However, you should limit the amount of HTML that translators must keep exactly correct, which means avoid including tags that wrap the entire translatable string, especially <p>.

    <p><&|/l, $button &>Do <em>not</em> click [_1]</&></p> # ok

    <&|/l, $button &><p>Do <em>not</em> click [_1]</p></&> # not ok

In a few places in RT we also pass HTML as parameters to loc() so that translators do not have to reproduce it exactly, and we can also change it more freely. For example:

    <&|/l,
        '<a href="http://www.gnu.org/licenses/gpl-2.0.html">',
        '</a>',
    &>Distributed under [_1]version 2 of the GNU GPL[_2].</&>

Translation Extraction

devel/tools/extract-message-catalog looks for loc("...") and <&|/l&>...</&> in our source code to pick out translatable strings, clean them up, and put them into share/po files. We use our .po files not only to populate Locale::Maketext's lexicons, but also to sync new translatable strings and translations with Launchpad. This Launchpad sync is typically done early during the freeze of RC releases to give our volunteer translators time to translate all the new strings which, because of the RC freeze, won't continue changing.

Translating Hard-coded Strings

Because loc() and /l are used to generate strings for human eyes, they generally must be used "close to the browser". These are directly in Mason templates, or in functions that return text that will be passed through Mason. However, in many places in RT we have hard-coded strings which need translations. For example, the $RIGHTS hash in lib/RT/Queue.pm maps rights' names (which must be translatable) to their descriptions (which also must be translatable). However, when we're declaring such structures, we do not want to translate them straight away. RT uses English internally, including in its web forms, so we do not want to localize rights' names except for display, otherwise things might break weirdly when you check if a user has the "Superusuario" right. Furthermore, when we're declaring such data structures at compile time, there is no current user to select which language to use for localization. Thus, we cannot call loc() when declaring $RIGHTS and other similar places.

For this reason, devel/tools/extract-message-catalog lets you denote translatable strings with comments. That's what the #loc_pair comments in the $RIGHTS hash in lib/RT/Queue.pm indicate. Since we have those comments, our toolchain will put the rights' names and descriptions into share/po files, which enables translation by our lovely volunteers. Later on, when RT displays information about rights in the web UI, we'll pass the right's name through loc, and Locale::Maketext will then be able to find our "Superusuario". So although we never used a literal loc("SuperUser"), we still get its effects thanks to the #loc_pair comments and using loc($RightName).

Translation Comment Markers

#loc_pair is used for declaring that both the key and value of a particular key => value pair are translatable. There are other markers that you can use.

#loc is used for declaring that a particular string is translatable. Its parsing is pretty strict so you can use it to declare that only the value of a particular key => value pair is translatable.

#loc_left_pair is used for declaring that the key of a particular key => value pair is translatable. This is of very limited usefulness.

#loc_right_pair does NOT exist. #loc works in such cases since its parser does not extend beyond the string at the end of a line. However, if the string is not at the end of the line, #loc{word} declares that the value associated with the key word (earlier on the same line) is to be loc'd. This is useful for inline hashes:

    # Note the string "baz" is to be loc'd
    foo => { bar => "baz", troz => "zort" },  # loc{bar}

String Extraction in Web Templates

Web templates

Templates should use the /l filtering component to call the localisation framework

The string Foo!

Should become <&|/l&>Foo!</&>

All newlines should be removed from localized strings, to make it easy to grep the codebase for strings to be localized

The string Foo Bar Baz

Should become <&|/l&>Foo Bar Baz</&>

Variable substitutions should be moved to Locale::MakeText format

The string Hello, <%$name %>

should become <&|/l, $name &>Hello, [_1]</&>

Multiple variables work just like single variables

The string You found <%$num%> tickets in queue <%$queue%>

should become <&|/l, $num, $queue &>You found [_1] tickets in queue [_2]</&>

When subcomponents are called in the middle of a phrase, they need to be escaped too:

The string <input type="submit" value="New ticket in">&nbsp<& /Elements/SelectNewTicketQueue&>

should become <&|/l, $m->scomp('/Elements/SelectNewTicketQueue')&><input type="submit" value="New ticket in">&nbsp;[_1]</&>

The string <& /Widgets/TitleBoxStart, width=> "40%", titleright => "RT $RT::VERSION for RT->Config->Get('rtname')", title => 'Login' &>

should become <& /Widgets/TitleBoxStart, width=> "40%", titleright => loc("RT [_1] for [_2]",$RT::VERSION, RT->Config->Get('rtname')), title => loc('Login'), &>

Library code

Within RT's core code, every module has a localization handle available through the 'loc' method:

The code return ( $id, "Queue created" );

should become return ( $id, $self->loc("Queue created") );

When returning or localizing a single string, the "extra" set of parenthesis () should be omitted.

The code return ("Subject changed to ". $self->Data );

should become return $self->loc( "Subject changed to [_1]", $self->Data );

It is important not to localize the names of rights or statuses within RT's core, as there is logic that depends on them as string identifiers. The proper place to localize these values is when they're presented for display in the web or commandline interfaces.

TESTING

Test suite

RT comes with a fairly complete test suite. The *-trunk and master branches are expected to always be passing all tests. While it is acceptable to break tests in an intermediate commit, a branch which does not pass tests will not be merged. Ideally, commits which fix a bug should also include a test case which fails before the fix and succeeds after.

To run RT's test suite, first set environment variables to a database user and password which can create and drop databases:

    export RT_DBA_USER=root
    export RT_DBA_PASSWORD=

You'll need to configure RT and make sure you have all the dependencies before running tests. To do this in place without installing:

    ./configure.ac --with-my-user-group --enable-layout=inplace --enable-developer
    make testdeps
    make fixdeps

Adjust the relevant database options as necessary if you want to test on PostgreSQL, Oracle, or SQLite. The default is MySQL/MariaDB.

To run the test suite:

    make test

If you want to run only a subset of the tests, you can use Perl's standard prove utility.

    prove -l t/api/ticket.t t/web/ticket_display.t

Parallel Tests

If you have multiple processors, you can run the test suite in parallel, which will be significantly faster:

    make test-parallel

If you want to run a subset of the tests in parallel, you can pass the -j flag to prove with the number of processes to use. Make sure to also set the RT_TEST_PARALLEL environment variable to a true value to tell RT you're running the tests in parallel.

    RT_TEST_PARALLEL=1 prove -l -j4 t/customfields/*.t

To run playwright tests sequentially and others in parallel:

    RT_TEST_PARALLEL=1 prove -l -j4 --rules='seq=t/playwright/*.t' --rules='par=**' t

Playwright Tests

Starting with version 6.0.3, RT uses Playwright for browser-based testing. Playwright allows you to test functionality with a virtual browser, making it possible to test behavior that happens only in the browser, specifically javascript features. This is also helpful for confirming htmx interactions. Selenium was used in previous versions.

Playwright tests are located in the t/playwright/ directory.

Prerequisites

You need Node.js and the Perl Playwright module installed:

    # On Mac, we had issues when installing via brew. Installing
    # nvm seems to be the current recommended approach.
    # See https://nodejs.org/en/download for details.

    # Install the Perl Playwright module
    cpanm Playwright

After installing the Perl module, install the browser drivers:

    npx playwright install firefox

You can also install other browsers (chromium, webkit) if needed.

Running Playwright Tests

Run all playwright tests:

    prove -l t/playwright/

Run a specific test:

    prove -lv t/playwright/ticket_create.t

Environment Variables

RT_TEST_PLAYWRIGHT_HEADLESS

Set to 0 to run browser in non-headless mode for debugging. Default: 1

    RT_TEST_PLAYWRIGHT_HEADLESS=0 prove -lv t/playwright/ticket_create.t

When set to 0, you can see the browser window and watch tests execute, which is helpful for debugging test failures or understanding page interactions.

RT_TEST_PLAYWRIGHT_BROWSER

Browser type to use: 'firefox', 'chromium', or 'webkit'. Default: firefox

    RT_TEST_PLAYWRIGHT_BROWSER=chromium prove -lv t/playwright/ticket_create.t

Firefox is the default as it has been tested most extensively with RT.

Crypt Tests

Tests for RT's GnuPG and SMIME features have some additional dependencies and environment variables to enable or disable tests.

To run the GPG tests, you need a working gpg executable of a version supported by GnuPG::Interface. If your environment isn't set up for this, you can set SKIP_GPG_TESTS to true to skip tests. The tests also check for gpg and don't run if it's not found.

Some of the SMIME tests access an internet resource to check certificates. This is disabled by default since some test systems don't have internet access. To run these tests, set RT_TEST_SMIME_REVOCATION to true.

Tests with Docker

RT has a Dockerfile that can be used to run the test suite for some configurations. To run it you need a system that can run docker natively or run the docker desktop client. Once installed, you can look in the .github/workflows/test-all.yml file for the set of commands to run to kick off the tests inside docker containers. Run all commands in the "Build RT" and "Run RT tests" sections, replacing any variables with appropriate values for your system.

You can replace $GITHUB_WORKSPACE with your current working directory if you are in a branch. Note that the docker commands in the configuration mount your directory inside the docker container. This means any stray files you have in your working directory will also be visible to the processes inside the docker container. This can cause test failures if unexpected files are there.

VERSION CONTROL

RT's source code is stored in a git repository. The RT source repository is available via git from GitHub; you can browse it at http://github.com/bestpractical/rt/ or obtain a local copy via:

    git clone git://github.com/bestpractical/rt.git

Branch Organization

The bleeding-edge development happens in the master branch. When a major release is anticipated, a "trunk" branch will be branched from this -- for example, 4.0-trunk. This will allow the trunk to stabilize while feature development continues on master. Additionally, as a release is impending for a particular series, a release engineering branch will be created, named, for example 4.0.0-releng.

New feature development should always be based off of the master branch. Branches to fix bugs should be based off of whichever trunk the bug was first found in. If you found the bug in your RT 4.0.0 install, you'd branch from 4.0-trunk.

Branches should be named based on the trunk they are branched from -- which is to say, the earliest branch they might be merged into. For example, a bugfix branched from 4.0-trunk might be named 4.0/fail-taint-mode-early. A feature branched from master when there exists a 4.0-trunk but no 4.2-trunk might be named 4.2/rename-LogToScreen. For consistency, branches should use dashes, not underscores, to separate words. Branches which are destined for 4.2, but which are branched from 4.0 (to provide for easy extraction as a 4.0 extension) should be named 4.2-on-4.0/branch-name.

Code Review

Branches should be reviewed by another developer before being merged. Reviewers should make sure that the branch accomplishes what it claims to, and does not introduce any unwanted behavior in doing so. Commit messages explain the why as much as the what of each commit, and not include extraneous changes.

Git Config Files

The RT repo has a .gitattributes file with a line that looks for a diff configuration for minified javascript. To help git generate readable diffs, you can use a configuration like this after installing the js-beautify utility.

[diff "minjs"] textconv = js-beautify cachetextconv = true

Git Quickstart

The process below describes how to get a copy of an RT repo, modify it, and submit your changes as a patch. You can also create an account on GitHub, fork RT, and submit a PR. GitHub has documentation on this process: https://help.github.com/articles/fork-a-repo/.

  1. You will first need to obtain a copy of git; this is accomplished via sudo yum install git in RedHat and derivatives, or sudo apt-get install git for Debian or Ubuntu.

  2. Next, obtain a copy of the RT source from git:

        git clone git://github.com/bestpractical/rt.git
    cd rt

  3. Configure git to know your name and email address; git uses these when it makes commits.

        git config user.email your.email@example.com
    git config user.name Examp L. Name

  4. Switch to the appropriate point to base your work on; this is generally origin/ followed by the major version, followed by -trunk. For example, if your bug was observed in version 3.8.9, you would choose origin/3.8-trunk; if it was in 4.0.0, you would choose origin/4.0-trunk. New features should be based on origin/master.

        git checkout --track origin/4.0-trunk

  5. Give your branch a name based on what you are attempting to accomplish. We suggest that branch names be lower-case and separate words with dashes, but this branch name is purely for your own reference.

        git branch -m gnupg-encryption

  6. Edit the source tree to make your changes. A few commands you may find useful in doing so are listed below.

    To see what files you have changed:

        git status

    To see a line-by-line list of changes:

        git diff

    To revert a file to the original version:

        git checkout path/to/file

    To revert only individual parts of a file:

        git checkout -p path/to/file

  7. Check that you have no extraneous changes using git diff, then commit your changes:

        git commit -a

    You will be prompted to type your commit message. The first line should be a short (< 80 character) summary of the changes, followed by a blank line, followed by a longer description, if necessary. The commit message should not simply restate the diff of which lines were added and subtracted, but should rather explain what those changes accomplish, and why they are desired.

    If your changes are easily split into multiple components, you may wish to split your changes into more than one commit; simply return to step 6 and repeat the with the next related change. If your changes are not related to each other, you should submit them separately; finish step 9, then start over from step 4.

  8. Save your commits to patch files:

        git format-patch @{u}

    This will print out the names of the files as it creates them.

  9. Attach these files to an email using your standard email client, and send it to rt-bugs@bestpractical.com. This will create a ticket in our public RT instance at https://issues.bestpractical.com.

If you have another bug or feature to implement, simply restart the process at step 4.

CODING PROCEDURE

This is for new programs, modules, specific APIs, or anything else.

Create a topic in RT Developers on the Forum

We may know of a better way to approach the problem, or know of an existing way to deal with it, or know someone else is working on it. This is mostly informal, but a fairly complete explanation for the need and use of the code should be provided.

Present specs in RT Developers

The complete proposed API should be submitted for discussion. For web and command-line programs, present the functionality and interface (op codes, command-line switches, etc.).

The best way to do this is to take the documentation portion of the boilerplate and fill it in. You can make changes later if necessary, but fill it in as much as you can.

Prepare for code review

When you are done, the code will undergo a code review by a member of the core team, or someone picked by the core team. This is not to belittle you (that's just a nice side effect), it is to make sure that you understand your code, that we understand your code, that it won't break other code, that it follows the documentation and existing proposal. It is to check for possible optimizations or better ways of doing it.

Note that all code is expected to follow the coding principles and style guide contained in this document.

Finish it up

After the code is done (possibly going through multiple code reviews), submit your updates as a pull request on GitHub. If you don't have a GitHub account, you can generate patches and send email to rt-bugs@bestpractical.com which will create a ticket in our public issue tracker at https://issues.bestpractical.com.

BUG REPORTS, PATCHES

Use rt-bugs@bestpractical.com for any bug that is not being fixed immediately. If it is not in RT, there is a good chance it will not be dealt with.

Send patches to rt-bugs@bestpractical.com, too. Use diff -u for patches.

SCHEMA DESIGN

RT uses a convention to denote the foreign key status in its tables. The rule of thumb is:

When it references to another table, always use the table name

For example, the Template field in the Scrips table refers to the Id of the same-named Template table.

Otherwise, always use the Id suffix

For example, the ObjectId field in the ACL table can refer to any object, so it has the Id suffix.

There are some legacy fields that did not follow this rule, namely ACL.PrincipalId, GroupMembers.GroupId and Attachments.TransactionId, but new tables are expected to be consistent.

EXTENDING RT CLASSES

The Overlay mechanism

RT's classes allow "overlay" methods to be placed into files named Filename_Vendor.pm and Filename_Local.pm. _Vendor is for 3rd-party vendor add-ons, while _Local is for site-local customizations.

These overlay files can contain new subs or subs to replace existing subs in this module.

Each of these files should begin with the line:

   no warnings qw(redefine);

so that perl does not kick and scream when you redefine a subroutine or variable in your overlay.

Some common ways that overlays are used:

Adding Methods

Create a file named Classname_Local.pm as appropriate (like User_Local.pm) in rt_base_dir/local/lib/RT/.

    no warnings qw(redefine);

    sub MyNewMethod {
        my $self = shift;
        ...
    }

    1;

Modifying Methods

Create a file with a _Local.pm suffix as appropriate in the same local/lib tree mentioned above. Copy the method you need to modify from the original RT version of the file and paste it into your local version. Then modify the code to behave the way you need it to.

When changing code in this way, make note of incoming values and especially return values since other code likely expects the existing method to return values a certain way.

When copying code for modification, do not place the _ImportOverlays call from the original at the bottom of your modified file.

    no warnings qw(redefine);

    sub ExistingMethod {
        ... existing RT code
        ... my special changes
        ... existing RT code
    }

    1;

Hooking Methods

Set up your local file the same way as in Modifying Methods described above. You'll need to save the original method call before you redefine it, and then call it at the head or tail of your code:

    no warnings qw(redefine);

    # This should be the same class we are overlaying here
    my $original_method = \&RT::Class::MethodToHook;

    sub MethodToHook {
        my $self = shift;
        ...
        ... my special code
        ...
        # Call the original method at the tail of our extra code:
        &$original_method( $args );
    }

    1;

Remember! If you modify existing core RT methods, you will need to update your local modifications when you upgrade the base RT code so that it matches.

WEB INTERFACE

Using TitleBox for Portlets

RT uses the TitleBox widget to create consistent portlet-style UI components throughout the interface. A titlebox renders as a Bootstrap card with a header (title) and body section. This section explains how to use titleboxes effectively, including integration with HTMX for lazy loading.

Basic Usage

The simplest titlebox wraps content in a card with a title:

    <&| /Widgets/TitleBox, title => loc('My Portlet'), class => 'my-portlet' &>
      <p>Content goes here</p>
    </&>

Common parameters:

title

The text displayed in the card header. Use loc() for internationalization.

class

CSS class(es) added to the outer titlebox div. Used for styling and targeting with CSS or JavaScript.

title_class

CSS class(es) added to the card header element.

hideable

Boolean (default 1). When true, displays a toggle to collapse/expand the portlet.

rolledup

Boolean (default 0). When true, the portlet starts in collapsed state.

titleright_raw

Raw HTML to display on the right side of the title bar, commonly used for action links or icons.

icons_ref

Reference to an array of hashes defining icon buttons in the title bar. Each hash can contain: icon_name, tooltip_text, icon_href, class, modal, and HTMX attributes.

content_body_class

CSS class(es) added to the inner card-body div. Useful for adjusting padding in special layouts (see "Card Body Spacing" below).

data

Hash reference of data attributes to add to the titlebox element.

Card Body Spacing

RT's titlebox uses Bootstrap's card component with customized padding. The card body has a default top padding of 0.5rem to create balanced spacing below the title.

For forms and other content using the LabeledValue template, RT's CSS automatically detects when the first visible element has the mt-2 class and removes the card body's top padding. This prevents double spacing since LabeledValue already includes top margin.

Important: This automatic adjustment means that adding mt-2 to your first element will not increase spacing from the title - instead, the CSS removes the card body padding to maintain consistent spacing. This is intentional behavior that ensures forms render correctly.

For special layouts that don't use the mt-2 pattern (such as multi-column layouts with their own spacing), you can explicitly control the padding:

    <&| /Widgets/TitleBox, title => loc('Custom Layout'),
        content_body_class => 'pt-0',
    &>
      <div class="row">
        <div class="col-6">...</div>
        <div class="col-6">...</div>
      </div>
    </&>

The CSS also automatically reduces padding for content starting with tables or accordions, which have their own internal spacing.

Using LabeledValue for Forms

The LabeledValue template provides consistent form field layout within titleboxes:

    <&| /Widgets/TitleBox, title => loc('Settings') &>
      <&| /Elements/LabeledValue, Label => loc('Name') &>
        <input type="text" class="form-control" name="Name" />
      </&>
      <&| /Elements/LabeledValue, Label => loc('Description') &>
        <textarea class="form-control" name="Description"></textarea>
      </&>
    </&>

Each LabeledValue creates a row with the mt-2 class, providing consistent vertical spacing between fields.

HTMX Integration for Lazy Loading

Titleboxes support HTMX-based lazy loading, where the card header renders immediately and the body content loads asynchronously. This improves initial page load time for pages with many portlets.

To enable lazy loading, use htmx_load with an htmx_get or htmx_post URL:

    <&| /Widgets/TitleBox,
        title => loc('Quick Create'),
        htmx_load => 1,
        htmx_get => RT->Config->Get('WebPath') . '/Views/Component/QuickCreate',
    &>
    </&>

When htmx_load is enabled:

For content under /Views/, the TitleBox component automatically detects the path and returns only the inner content without the card wrapper. This allows the same Mason component to work both in the initial page render and as an HTMX response.

To force full titlebox rendering even for a /Views/ path (useful for components that might be loaded in different contexts), explicitly set htmx_load => 0:

    <&| /Widgets/TitleBox, title => loc('My Content'), htmx_load => 0 &>
      ...
    </&>

Refreshing Content with HTMX

Titlebox content can refresh in response to custom HTMX events. Add hx-trigger to an inner div to listen for events:

    <&| /Widgets/TitleBox, title => loc('Ticket Info') &>
      <div hx-trigger="ticketBasicsChanged from:body"
           hx-get="/Views/Ticket/ShowBasics?id=123"
           hx-swap="innerHTML">
        <& /Ticket/Elements/ShowBasics, ... &>
      </div>
    </&>

When JavaScript dispatches a ticketBasicsChanged event on the body, HTMX automatically refreshes the inner div content.

Example: Complete Portlet with Inline Edit

Here's a more complete example showing a portlet with inline editing capability:

    <&| /Widgets/TitleBox,
        title => loc('The Basics'),
        class => 'ticket-info-basics',
        titleright_raw => $can_modify
            ? $m->scomp('/Elements/InlineEditLink')
            : '',
        data => { 'inline-edit-behavior' => 'link' },
    &>
      <div class="inline-edit-display"
           hx-trigger="ticketBasicsChanged from:body"
           hx-get="/Views/Ticket/ShowBasics?id=<% $id %>">
        <& /Ticket/Elements/ShowBasics, ... &>
      </div>
      <form hx-post="/Helpers/TicketUpdate"
            hx-swap="none"
            class="inline-edit">
        <input type="hidden" name="id" value="<% $id %>" />
        <& /Ticket/Elements/EditBasics, ... &>
        <div class="row mt-2">
          <div class="col-12 text-end">
            <input type="submit" class="btn btn-primary"
                   value="<&|/l&>Save</&>" />
          </div>
        </div>
      </form>
    </&>

TO DO

Talk about DBIx::SearchBuilder

Talk about mason component style cascading style sheets

Talk about adding a new translation

Talk more about logging

← Back to index

Security contact information
© 2002-2026 Best Practical Solutions LLC.