Originally submitted to Google Code by @pekkaklarck on 31 May 2008

• explain different templates briefly
• few more words why attd-templates are different
  • only two columns in test case tabe
  • has only test case and setting tables (check that no other tables)
  • can use Test/Suite Precondition/Postcondition
• mention that templates can be customized freely
  • contents of the first cell used for recognizing the table
  • refer to user guide about other syntax

Originally submitted to Google Code by @pekkaklarck on 31 May 2008

1. README is needed
2. At least install.sh should be converted to Python. That way same script
   could be used also in Windows and there would be no need for separate
   win_uninstall.sh (which only work on Cygwin)

Originally submitted to Google Code by @pekkaklarck on 30 May 2008

• Start Time, End Time, Elapsed Time should be combined to one row
• Tests' criticality (Yes/No) should be shown with status (or somewhere else)
• Perhaps combine status and message
• Perhaps show suites' statistics message on one line

Originally submitted to Google Code by @pekkaklarck on 28 Jun 2008

Currently the version in the text right below the title is hard-coded, and
when a ug distribution is created version is given from command line. It
would be better to:

1. Get version number automatically from version.py similarly as setup.py
   gets it. This is easy.

2. Set version dynamically also to the generated HTML. This may be easy if
   reStructuredText allows given substitutions from the command line, but
   editing the source file before HTML is generated is not much harder either.

All this makes it easier to have consistent version numbers everywhere.

Originally submitted to Google Code by @pekkaklarck on 15 Jun 2008

The current system doesn't really work.

• Not good to have same files multiple times in the version control
• Source distribution grows big
• doc/userguide/rfug has .svn directory which is included if rfug zipped

I think something like this could work much better:

1. Create RobotFrameworkUserGuide.html directly to doc/userguide.

• This file would be in version control
• Other docs, images, etc. linked so that they are not moved anywhere
  • no duplication
  • relative links ought to work both online and in source distro
• Created with "python ug2html.py"

2. Have a separate commend for creating a stand-alone distribution

• For example "python ug2html.py dist"
• Needed actions:
  • first create the normal doc
  • create new rfug directory (delete possible existing one)
  • copy doc/libraries to rfug/libraries (without .svn dir)
  • create rfug/tools and copy tool docs there
  • create rfug/images
  • copy RobotFrameworkUserGuide.html to rfug
  • adjust lib and tool links so that they point to docs under rfug
  • copy used images to rfug/images and adjust paths accordingly
• rfug would not be in version control (no .svn dirs)
• could have "python ug2html.py zip" to also zip rfug automatically
• also "python ug2html.py help" or similar needed

Originally submitted to Google Code by @pekkaklarck on 31 May 2008

The README.txt file in the root is not that good. It has way too technical
introduction (don't now should it mention e.g. 'pybot' and 'jybot' at all)
and doesn't really explain that the framework actually does.

A good introduction is needed also in the user guide and here in Google
Code. Most likely the same text can be used at least to some extend on all
places. We need to sit down (preferably with few beers) and come up with
some good "marketing text".

Another problem is that license and copyright are not mentioned at all in
the README.

Originally submitted to Google Code by @pekkaklarck on 31 May 2008

This issue has two separate but strongly related "sub issues":

A. User guide should mention the copyright/license used with the framework.

B. User guide itself should use the same copyright/license.

I propose that we do following:

1. Add "Copyright and license" section to "Getting started". The section
   doesn't need much more than the normal copyright statement in the beginning
   of every source code file.

2. Add a short copyright notice in the beginning of the UG. Perhaps just
   have "(C) Nokia Siemens Networks 2008", mention Apache 2 license, and link
   to "Copyright and license" section.

3. Add full license text as an appendix. Apache 2 license is only 202 lines
   so it doesn't increase the size that much. Appendix can be linked from the
   "Copyright and license" section.

4. Perhaps add normal copyright notice to every reST source file. I'm not
   sure is that necessary, but other source files have them and it's not that
   big a task.

Originally submitted to Google Code by heiko.selber on 22 Jul 2008

Currently the documentation of the Robot Framework doesn't make it very
easy for new users to create test cases.

Therefore, it would be desirable to have a collection of very basic test
case examples, each of which shows the principle of one of Robot's features.

Here is a quick list of candidates:

helloworld.html: run one very small test case
numbers.html: simple number comparisons, e.g. 2*3=6 or sin(pi)=0 etc.
regexp.html: regexp matching
conditions.html: 'if' statements and 'for' loops
variable.html: show the usage of various variables
run.html: run a binary with command line arguments and evaluate its return
codes
multitasking.html: run processes or threads in the background; examine
their status; connect to them; interrupt them
GUI.html: execute actions on a SWING GUI

Originally submitted to Google Code by @pekkaklarck on 31 May 2008

Better to go through all the issues together and

1. resolve those that are fixed
2. just close possible invalid ones
3. move generic issues here, and add a link to the original issue
   4 ) keep domain specific or otherwise internal issues where they are

Originally submitted to Google Code by @yanne on 10 Jul 2008

This is from Telnet Library documentation of 'Set Timeout':

"""
Sets the timeout used in read operations to the given value.

'timeout' is given in Robot Framework's time format
(e.g. 1 minute 20 seconds).

The read operations wait for this time before starting to read
from the output. To run operations that take a long time to generate
their complete output, this timeout must be set accordingly.
"""

The third chapter is clearly wrong, since the read operations start
immediately after command is written and last until either the expected
output is gotten or timeout occurs.

Probably this should also list the operation the timout affects, since not
all read operations use the timeout.

Originally submitted to Google Code by @jrantanen on 27 Jun 2008

Possible to import libraries using physical path

It is possible to import libraries using a physical path to them. This path
is considered relative to the directory where current test data file is
situated similarly as paths to resource and variable files.

If the library is a file, the path to it must contain extension. For Python
libraries the extension is naturally ".py" and for Java libraries it can
either be ".class" or ".java" but the class file must of course be
available. Importing Java libraries in jar files is not possible using this
mechanism. If Python library is implemented as a directory, the path to it
must have a trailing "/" separate the usage from normal importing.
Following examples demonstrate these different usages.

| Library | PythonLib.py |
| Library | /absolute/path/JavaLib.java |
| Libarry | relative/path/PythonDirLib/ |

Originally submitted to Google Code by DeonWu on 2 Jul 2008

What steps will reproduce the problem?

1. Append To List|${list}|${list_item}
2. raise a Error if the ${list_item} is not a string object.
   "TypeError: cannot concatenate 'str' and 'list' objects"

What version of the product are you using? On what operating system?
Robot 1.8.8 (Python 2.5.2 on win32)

Please provide any additional information below.

def append_to_list(self, L, *values):
    """Adds 'values' to the end of L.

    Example:
    | Append To List | ${L1} | xxx |   |   |
    | Append To List | ${L2} | x | y | z |
    =>
    ${L1} == ['a', 'xxx']
    ${L2} == ['a', 'b', 'x', 'y', 'z']
    """
    for value in values:
        print 'Appending value ' + value
        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 
        L.append(value

Originally submitted to Google Code by @pekkaklarck on 16 Jun 2008

• Better/more "marketing material"
• Screenshots of reports, logs, data
• More overview information about the architecture etc.
• Would "Introduction" be a better name for the chapter?

Originally submitted to Google Code by @pekkaklarck on 23 Jul 2008

None of the keywords in Collections library log what they are doing *). It
might be valuable information sometimes, but then again lists and dicts can
be really big and logging them would increase the log size. One possibility
is logging only changes, and other is logging everything on DEBUG level. I
think the latter approach is better, especially if there is a mechanism to
alter the logging level (like keyword 'Set Collections Log Level' or/and
argument to constructor).

Juha, you have more experience about this library. How important logging is
in your opinion and how should it be handled?

*) 'Append To List' used log in 2.0, but there was a bug (see issue 45 ) and
logging was removed in 2.0.1.

Originally submitted to Google Code by @pekkaklarck on 31 May 2008

Backlog meaning post-it notes on the wall. =)

Originally submitted to Google Code by @pekkaklarck on 30 May 2008

Need to go through once more, add correct download links, etc.

Originally submitted to Google Code by @pekkaklarck on 31 May 2008

High-level outline is ready but otherwise this is under construction. This
is probably the biggest task lest before the official release.

Originally submitted to Google Code by @pekkaklarck on 31 May 2008

How should different tools be documented? They all should at least have
--help, but that is not enough with tools needing more detailed
instructions, screenshots, etc.

My current feeling is that every tool should have it's own documentation
written using reST. These docs can then be distributed with the tool, and
also attached to the user guide similarly as library docs. For simple tools
it might be enough, and even recommended, to have the --help text already
in reST.

One open question is that where to keep those docs. I think it's better to
keep them under tools/ in the same directory where the tool itself is, but
creating a new doc/tools/ is an option too.

Originally submitted to Google Code by @pekkaklarck on 16 Jun 2008

• Need some "marketing material" and screenshots - can use same as in the
  user guide
• Links to the user guide, standard libraries, tools, ...
• Users and code mailing lists

Originally submitted to Google Code by @yanne on 4 Jun 2008

The eclipse settings files could be added to
version control just to provide them, since
they contain nothing too specific.

Originally submitted to Google Code by @pekkaklarck on 16 Jun 2008

Elisabeth has been working with Getting started guide. Demos on it are
based on her old examples, and they need to be ported from Ruby to Python
so that having only Python installed is enough to run them.

Originally submitted to Google Code by @pekkaklarck on 31 May 2008

• Includes a better system for setting version
• Version should also include alpha/beta/rc information
• Needed scripts should be written in Python to support all platforms
• Need to write instructions how to make releases at the same time (how to
  use branches, tags, ...)

Originally submitted to Google Code by sampeson on 20 Jun 2008

In example 1 for loop uses ${animal} variable but then inside loop logs
${var} (should use ${animal} otherwise example does not make sense).

Originally submitted to Google Code by @yanne on 26 Jun 2008

Telnet tests need to refactored so that they can be easily executed using
arbitrary server and credentials.

Originally submitted to Google Code by @pekkaklarck on 25 Jun 2008

PyPI = Python Package Index = http://pypi.python.org

It seems to be easiest to just submit PKG-INFO file which can be created
using setup.py. Before that we need to update the metadata in setup.py,
since currently it contains only minimal information.

Originally submitted to Google Code by @pekkaklarck on 15 Jun 2008

After running tests I got this:

peke@⁠omppu:~/workspace/robotframework$ svn stat
? atest/testdata/public_api/output

Two possibilities:

1. Remove the directory in a suitable teardown
2. Create it into atest/results/<timestamp> in the first place. If I
   remember correctly, this directory is in ${OUTDIR} variable on running side.

Adding the directory to svn:ignore is not a good solution.

Originally submitted to Google Code by @pekkaklarck on 31 May 2008

• explain different templates briefly
• few more words why attd-templates are different
  • only two columns in test case tabe
  • has only test case and setting tables (check that no other tables)
  • can use Test/Suite Precondition/Postcondition
• mention that templates can be customized freely
  • contents of the first cell used for recognizing the table
  • refer to user guide about other syntax

Originally submitted to Google Code by @yanne on 12 Jun 2008

We should implement RPM-package distribution of Robot Framework

Originally submitted to Google Code by @yanne on 30 May 2008

Report file should have a tree like structure
with expand/collapse buttons similar to the log file to make
visualization of hierarchical structure easier.

Originally submitted to Google Code by @pekkaklarck on 31 May 2008

Most of the appendices still need a review/rewrite. They can be excluded
from the first release if we run out of time, but it would be much better
to get them to a good shape too.

Originally submitted to Google Code by miso.korkiakoski on 2 Jun 2008

What steps will reproduce the problem?

1. Checkout sources
2. install using 'python setup.py install"
3. run acceptance tests with 'atest/run_atests.py python --splitoutputs 2 atest/robot'

What is the expected output? What do you see instead?
PASS on all critical tests, instead I see 3 failed critical tests. Possibly a OS issue?

What version of the product are you using? On what operating system?
svn checkout taken on the 2nd of June. Running on OS X Leopard (10.5.3 update)

Please provide any additional information below.
Test reports attached.

• Miso

Originally submitted to Google Code by @pekkaklarck on 15 Jun 2008

Some tests currently fail if all tests are executed with "--include
jybotNOTpybot". This is due to those tests being dependent on "pybot only"
tests, and the fix is to make those tests independent.

Can't test which exact tests fail, since I don't have Jython installed on
this machine.

Originally submitted to Google Code by @pekkaklarck on 16 Jun 2008

User guide still has some TODOs that need to be resolved. Related to this,
there are still some links to internal wiki and other resources.

Originally submitted to Google Code by @yanne on 4 Jun 2008

The PACKACKING.txt should contain information about
how to create RPM packages of Robot Framework.

Originally submitted to Google Code by @yanne on 26 Jun 2008

• Add a zip source distribution (no need to be featured)
• Test Windows installers created on Linux
  • if there are problems, package.py should prevent creating final Windows
    releases on non-Windows systems
• package.py --> packaging.py?
• Script for creating tag, updating version number and running wiki update
  script
• package.py should be modified so that it is possible to not update
  version when creating packages
• ug2html need to have 'archive' option
  • zip vs tar.gz?
• automatically upload packages to download page

Originally submitted to Google Code by @pekkaklarck on 2 Jul 2008

Separate user guide download package is mainly targeted for Windows users,
since user guide is included in the normal source distribution used on
other platforms. Native Windows can't even open tar.gz packages (e.g.
WinZip can but it is not installed by default), so it's probably better
that we package the user guide as zip in the future.

Originally submitted to Google Code by @yanne on 4 Jun 2008

INSTALL.txt should contain a working URL where distributions
could be downloaded from.

Originally submitted to Google Code by @pekkaklarck on 11 Jun 2008

Current screenshots in Output files section are not too good.

• They have totally dummy data. Better screenshot could be taken for
  example from SeleniumLibrary demos - they could even be linked from
  Demonstration section.
• They are created using an old version of the framework (seems to be 1.7).
  Both report and log look a bit different (=better) now and have some some
  new functionality that should be shown.
• There should be both green and red reports (only latter currently).
• Better screenshots can also be shown in wiki, possible getting started
  guide, etc.

Originally submitted to Google Code by DeonWu on 6 Jul 2008

What is the expected output? What do you see instead?

1. A Debugger window can track the testcase step by step.
2. add "--debug gui" on pybot command line, the pybot runinto a debug mode.
   And a Debuger window will pop-up.

Please provide any additional information below.
The patch for robot coding.

C:\USERS_work\robotframework-read-only\src\robot>svn di

Index: conf/settings.py

--- conf/settings.py (revision 369)
+++ conf/settings.py (working copy)
@⁠@⁠ -55,7 +55,8 @⁠@⁠
'TagStatLink' : ('tagstatlink', []),
'LogLevel' : ('loglevel', 'INFO'),
'MonitorWidth' : ('monitorwidth', 78),

•              'MonitorColors'    : ('monitorcolors', 'ON') }
  

•              'MonitorColors'    : ('monitorcolors', 'ON'),
  

•              'DebugMode'        : ('debug', None) }
  

  _deprecated = { 'colormonitor' : 'monitorcolors' }

Index: init.py

--- init.py (revision 369)
+++ init.py (working copy)
@⁠@⁠ -28,6 +28,7 @⁠@⁠
from serializing import RobotTestOutput, RebotTestOutput, SplitIndexTestOutput
from errors import DataError, INFO_PRINTED, DATA_ERROR, FRAMEWORK_ERROR
from variables import init_global_variables
+from debuger import TestSuiteDebuger
import utils

version = utils.version
@⁠@⁠ -80,6 +81,10 @⁠@⁠
pybot --log mylog.html /path/to/tests.html /path/to/tests2.html
"""
settings = RobotSettings(options)

• debuger = None
• if settings['DebugMode'] is not None:

•    debuger = TestSuiteDebuger(settings)
  

•    debuger.start()
  

  output = Output(settings)
  settings.report_errors(output.syslog)
  init_global_variables(settings, output.syslog)
  @⁠@⁠ -98,6 +103,8 @⁠@⁠
  testoutput = RebotTestOutput(datasources, settings, output.syslog)
  testoutput.serialize(settings, output.syslog)
  output.close2()
• if debuger is not None:

•    debuger.close()
  

  return suite

Index: runner.py

--- runner.py (revision 369)
+++ runner.py (working copy)
@⁠@⁠ -239,6 +239,7 @⁠@⁠
| # This is a comment line
| my_tests.html
| path/to/test/directory/

• -X --debug mode Debug mode(remote, gui)
  -h -? --help Print usage instructions.
  --version Print version information.

OTHER:

1. It's simply and ugly, because it major used to verify the feasibility of
   idea.
2. About the debugger window:
   a. you can add some breakpoints in the testcase at the right plane of
   debuger window.
   1. the breakpoints should a keyword name, and the wildcard is
   supported. The testcase will stopped when it matched the breakpoint keyword
   name.
   b. the debuger tools:
   1. Run -- the testcase will run with no pause until a breakpoint is
   mathced.
   2. Go into -- To execute a keyword, and stoped before next keyword
   executing, track into if the next keyword is a userkeyword.
   3. Go over -- To execute a keyword, and stoped before next same
   level keyword.
   4. Go Result -- To execute a keyword, and stoped before the current
   keyword return. (Used for watch the return result of current keyword,NOT
   SUPPORTED in currently.)
   c. the Varibles window will shoud the variabe name and value of current
   keyword.
   d. The Console window will list all the executed keywords, and the last
   one is the current paused keyword.
   e. The Call stack will list the caller relationship of userkeyword.

Originally submitted to Google Code by @pekkaklarck on 2 Jul 2008

User guide ought to have an appendix about templates with links to actual
template files. Generated user guide package should also include templates
similarly as it has e.g. library docs.

Originally submitted to Google Code by metalzong on 6 Jul 2008

Following sample shows an inconvenient situation - to call different
keywords with same signatory based on some condition - and the potential
solution:

UK section:

| My UK A | Log | A |
| My UK B | Log | B |
| My UK C | Log | C |

KW invoking without variable replacement:

| ${uk} | Set Variable | A |
.
.
.
| Run Keyword If | '${uk}' == 'A' | My UK A |
| Run Keyword If | '${uk}' == 'B' | My UK B |
| Run Keyword If | '${uk}' == 'C' | My UK C |

KW invoking with variable replacement:

| ${uk} | Set Variable | A |
.
.
.
| My UK ${uk} |

$ diff -c robot-1.8.8/src/robot/running/keywords.py
robot-1.8.8.bak/src/robot/running/keywords.py
*** robot-1.8.8/src/robot/running/keywords.py Mon Jul 7 11:10:07 2008
--- robot-1.8.8.bak/src/robot/running/keywords.py Tue Jun 10 17:04:58
2008

*** 25,32 ****
self.handler_name = name

  def run(self, output, namespace):

•     if str.handler_name.count("$"):
  

•         self.handler_name = Keyword("Set Variable",
  

  self.handler_name).run(out, namespace)
  handler = namespace.get_handler(self.handler_name)
  if handler.type == 'user':
  handler.init_user_keyword(namespace.variables)
  --- 25,30 ----

Originally submitted to Google Code by @pekkaklarck on 16 Jun 2008

Link targets for test cases and suites in a log file are created so that it
has the test case name prefixed with first letters of parent suite names.
In most cases this creates unique target names, but that is not guaranteed.
For example when running same test cases multiple times with different
variables test cases names are always the same, so duplicate targets are
created if parent suites start with the same letter.

An easy solutions is using the full name of the test case or suite. This
name includes all parent suite names and it is unique (at least in all
meaningful cases). The problem is that these names are quite long, but
probably they don't make log files so much bigger that it would be a problem.

Originally submitted to Google Code by @rgnyman on 3 Jun 2008

> Is it possible to take string handling and result verification
> libs to Robot team's responsibility?
> These are very generic functionalities.

Originally submitted to Google Code by @pekkaklarck on 30 May 2008

This should always, also after using Rebot, be the path to the original
test data file or directory.

Originally submitted to Google Code by @yanne on 17 Jun 2008

Running Telnet tests requires a Telnet server to be available.
The best solution would be a programmatic Telnet server that could be
started from the tests.

Originally submitted to Google Code by @pekkaklarck on 23 Jun 2008

It should be possible to package Robot Framework into a JAR package. That
way installing would only include dropping the package into suitable place
and Python would not be needed for that. Alternatively also Jython could be
included in the package so having only Java would be enough.

Originally submitted to Google Code by @pekkaklarck on 10 Jun 2008

We need the generated UG in version control to be able to link to it from
wiki, and to get it to the source distribution. The biggest open issue
related to this is how to handle linked/included content like images and
lib/tool docs.

Originally submitted to Google Code by @pekkaklarck on 16 Jun 2008

Need to add a welcome message to all our mailing lists. The message should
contain a one sentence introduction, link to http://robotframework.org and
links to other mailing lists. The lists we currently have are:

http://groups.google.com/group/robotframework-devel
http://groups.google.com/group/robotframework-users
http://groups.google.com/group/robotframework-commit

Originally submitted to Google Code by @yanne on 26 Jun 2008

Currently easy_install is able to download and install Robot Framework,
but post install script is not executed and thus the runner scripts (pybot,
jybot and rebot) don't work properly. After modifying runner scripts
manually, the installation seems to work correctly

Fixing this probably requires changing the mechanism how runner scripts are
created.

Originally submitted to Google Code by @pekkaklarck on 31 May 2008

Our standard libraries (BuiltIn, OperatingSystem, ...) must be documented
in wiki. Keyword docs can of course stay in separate HTML files linked from
wiki, but a page where all standard libraries are briefly described (and kw
docs linked from) is still needed. That page can have pretty much same
content as "Standard libraries" section in the user guide.

We probably don't need separate wiki pages for libraries, but that depends
slightly on how good links we get to each library docs. I want to be able
to refer to standard libraries in general and also each standard library
with a clear and relatively short url.

We also need another wiki page for external libraries. This page should
also have a short description of each library (only SeleniumLibrary in the
beginning) and link to library's project page.