Add missing links.

[docutils.git] / BUGS.txt

================
 Docutils_ Bugs
================

:Author: David Goodger; open to all Docutils developers
:Contact: goodger@python.org
:Date: $Date$
:Revision: $Revision$
:Copyright: This document has been placed in the public domain.

.. _Docutils: http://docutils.sourceforge.net/


Bugs in Docutils?!?  Yes, we do have a few.  Some are old-timers that
tend to stay in the shadows and don't bother anybody.  Once in a while
new bugs are born.  From time to time some bugs (new and old) crawl
out into the light and must be dealt with.  Icky.

This document describes how to report a bug, and lists known bugs.

.. contents::


How To Report A Bug
===================

If you think you've discovered a bug, please read through these
guidelines before reporting it.

First, make sure it's a new bug:

* Please check the list of `known bugs`_ below and the `SourceForge
  Bug Tracker`_ to see if it has already been reported.

* Are you using the very latest version of Docutils?  The bug may have
  already been fixed.  Please get the latest version of Docutils from
  the repository_ or from the current snapshot_ and check again.  Even
  if your bug has not been fixed, others probably have, and you're
  better off with the most up-to-date code.

  If you don't have time to check the latest snapshot, please report
  the bug anyway.  We'd rather tell you that it's already fixed than
  miss reports of unfixed bugs.

* If Docutils does not behave the way you expect, look in the
  documentation_ (don't forget the FAQ_!) and `mailing list archives`_
  for evidence that it should behave the way you expect.

If you're not sure, please ask on the Docutils-users_ mailing list
first.

---------------------------------------------------------------------

If it's a new bug, the most important thing you can do is to write a
simple description and a recipe that reproduces the bug.  Try to
create a `minimal example`_ that demonstrates the bug.  The easier you
make it to understand and track down the bug, the more likely a fix
will be.

.. sidebar:: minimal example

  .. _minimal example:

  A minimal example is an example which is as small as possible. These
  examples are much easier to understand than long examples.

  To construct an example which is as small as possible, the rule is
  quite simple: *remove anything which is not necessary*.

  See also: `LaTeX FAQ`__, `Lilypond Documentation`__, minimalbeispiel.de__

  __ http://www.tex.ac.uk/cgi-bin/texfaq2html?label=minxampl
  __ http://lilypond.org/doc/v2.9/Documentation/user/lilypond/Minimal-examples
  __ http://www.minimalbeispiel.de/mini-en.html

Now you're ready to write the bug report.  Please include:

* A clear description of the bug.  Describe how you expected Docutils
  to behave, and contrast that with how it actually behaved.  While
  the bug may seem obvious to you, it may not be so obvious to someone
  else, so it's best to avoid a guessing game.

* A complete description of the environment in which you reproduced
  the bug:

  - Your operating system & version.
  - The version of Python (``python -V``).
  - The version of Docutils (use the "-V" option to most Docutils
    front-end tools).
  - Any private modifications you made to Docutils.
  - Anything else that could possibly be relevant.  Err on the side
    of too much information, rather than too little.

* A literal transcript of the *exact* command you ran, and the *exact*
  output.  Use the "--traceback" option to get a complete picture.

* The exact input and output files.  Create a `minimal example`_
  of the failing behaviour — it is better to attach complete files
  to your bug report than to include just a summary or excerpt.

* If you also want to include speculation as to the cause, and even a
  patch to fix the bug, that would be great!

The best place to send your bug report is to the `SourceForge Bug
Tracker`_.  That way, it won't be misplaced or forgotten.  In fact, an
open bug report on SourceForge is a constant irritant that begs to be
squashed.

Thank you!

(This section was inspired by the `Subversion project's`__ BUGS__
file.)

__ http://subversion.tigris.org/
__ http://svn.collab.net/viewcvs/svn/trunk/BUGS?view=markup

.. _CVS: http://sourceforge.net/cvs/?group_id=38414
.. _repository: docs/dev/repository.html
.. _snapshot: http://docutils.sourceforge.net/#download
.. _documentation: docs/
.. _FAQ: FAQ.html
.. _mailing list archives: http://docutils.sf.net/#mailing-lists
.. _Docutils-users: docs/user/mailing-lists.html#docutils-users
.. _SourceForge Bug Tracker:
   http://sourceforge.net/tracker/?group_id=38414&atid=422030


Known Bugs
==========

Also see the `SourceForge Bug Tracker`_.

* The HTML writer generates invalid XHTML for _`centered images with
  targets`::

      .. image:: example.png
         :align: center
         :target: http://docutils.sourceforge.net/

  This results in ``<a><div><img /></div></a>``, which is invalid.

* .. _error reporting:

  Calling rst2s5.py with a non-existent theme (``--theme
  does_not_exist``) or a non-existent language (``--language zz``)
  causes exceptions.  Such errors should be handled more gracefully.

* The "stylesheet" setting (a URL, to be used verbatim) should be
  allowed to be combined with "embed_stylesheet".  The stylesheet data
  should be read in using urllib.  There was an assumption that a
  stylesheet to be embedded should exist as a file on the local
  system, and only the "stylesheet_path" setting should be used.

* ``utils.relative_path()`` sometimes returns absolute _`paths on
  Windows` (like ``C:/test/foo.css``) where it could have chosen a
  relative path.

  Furthermore, absolute pathnames are inserted verbatim, like
  ``href="C:/test/foo.css"`` instead of
  ``href="file:///C:/test/foo.css"``.

  For details, see `this posting by Alan G. Isaac
  <http://article.gmane.org/gmane.text.docutils.user/1569>`_.

* _`Line numbers` and "source" in system messages are inconsistent.

  - In text inserted by the "include" directive, errors are often not
    reported with the correct "source" or "line" numbers.

    This is partially fixed in the commits from 2009-09-25 and 2009-10-28.
    The test in test/test_parsers/test_rst/test_directives/test_include.py
    works, but there are still plenty of system messages pointing to the
    wrong spot.

  - Perhaps all Reporter calls need "source" and "line" keyword arguments?

    Alternatively, give `document.reporter` access to the state machine
    instance `document.statemachine` and extract the "source" and "line"
    info from `statemachine.input_lines`.
    Except for special cases, there is even no need to call with "line":
    During parsing, `document.statemachine` knows the current line number.
    For system messages generated after the parsing is completed (i.e. by
    transforms or the writer) "line" info must be present in the doctree
    elements.

  - Elements' .line assignments should be checked.  (Assign to .source
    too?  Add a set_info method?  To what?)

    The "source" (and line number in the source) could either be added
    explicitely to the elements or determined from the “raw” line number by
    `document.statemachine.get_source_spot`.

  - Some line numbers in elements are not being set properly
    (explicitly), just implicitly/automatically.  See rev. 1.74 of
    docutils/parsers/rst/states.py for an example of how to set.

  - The line numbers of definition list items are wrong::

        $ rst2pseudoxml.py --expose-internal-attribute line
        1
          2
          3

        5
          6
          7

        <document source="<stdin>">
            <definition_list>
                <definition_list_item internal:line="3">
                    <term>
                        1
                    <definition>
                        <paragraph internal:line="2">
                            2
                            3
                <definition_list_item internal:line="6">
                    <term>
                        5
                    <definition>
                        <paragraph internal:line="6">
                            6
                            7

* .. _none source:

  Quite a few nodes are getting a "None" source attribute as well.  In
  particular, see the bodies of definition lists.

* Footnote label "5" should be "4" when processing the following
  input::

      ref [#abc]_ [#]_ [1]_ [#4]_

      .. [#abc] footnote
      .. [#] two
      .. [1] one
      .. [#4] four

  Output::

      <document source="<stdin>">
          <paragraph>
              ref
              <footnote_reference auto="1" ids="id1" refid="abc">
                  2

              <footnote_reference auto="1" ids="id2" refid="id5">
                  3

              <footnote_reference ids="id3" refid="id6">
                  1

              <footnote_reference auto="1" ids="id4" refid="id7">
                  5
          <footnote auto="1" backrefs="id1" ids="abc" names="abc">
              <label>
                  2
              <paragraph>
                  footnote
          <footnote auto="1" backrefs="id2" ids="id5" names="3">
              <label>
                  3
              <paragraph>
                  two
          <footnote backrefs="id3" ids="id6" names="1">
              <label>
                  1
              <paragraph>
                  one
          <footnote auto="1" backrefs="id4" ids="id7" names="4">
              <label>
                  5
              <paragraph>
                  four

* IDs are based on names.  Explicit hyperlink targets have priority
  over implicit targets.  But if an explicit target comes after an
  implicit target with the same name, the ID of the first (implicit)
  target remains based on the implicit name.  Since HTML fragment
  identifiers based on the IDs, the first target keeps the name.  For
  example::

      .. contents::

      Section
      =======

      .. _contents:

      Subsection
      ----------

      text with a reference to contents_ and section_

      .. _section:

      This paragraph is explicitly targeted with the name "section".

  When processed to HTML, the 2 internal hyperlinks (to "contents" &
  "section") will work fine, but hyperlinks from outside the document
  using ``href="...#contents"`` and ``href="...#section"`` won't work.
  Such external links will connect to the implicit targets (table of
  contents and "Section" title) instead of the explicit targets
  ("Subsection" title and last paragraph).

  Hyperlink targets with duplicate names should be assigned new IDs
  unrelated to the target names (i.e., "id"-prefix serial IDs).

* The "contents" ID of the local table of contents in
  ``test/functional/expected/standalone_rst_pseudoxml.txt`` is lost in
  the HTML output at
  ``test/functional/expected/standalone_rst_html4css1.html``.

* _`Blank first columns` in simple tables with explicit row separators
  silently swallow their input.  They should at least produce system
  error messages.  But, with explicit row separators, the meaning is
  unambiguous and ought to be supported::

      ==============  ==========
      Table with row  separators
      ==============  ==========
                      and blank
      --------------  ----------
                      entries
      --------------  ----------
                      in first
      --------------  ----------
                      columns.
      ==============  ==========

  Added a commented-out test case to
  test/test_parsers/test_rst/test_SimpleTableParser.py.

* _`Footnote references with hyperlink targets` cause a possibly
  invalid node tree and make the HTML writer crash::

      $ rst2pseudoxml.py
      [1]_

      .. _1: URI
      <document source="<stdin>">
          <paragraph>
              <footnote_reference ids="id1" refuri="URI">
                  1
          <target ids="id2" names="1" refuri="URI">

* Anonymous references have "name" attributes.  Should they?  Are they
  used?  See ``test/test_parsers/test_rst/test_inline_markup.py``.

* <reference> elements have a "name" attribute, not "names".  The
  attribute should be "names"; this is an inconsistency.

\f
..
   Local Variables:
   mode: indented-text
   indent-tabs-mode: nil
   sentence-end-double-space: t
   fill-column: 70
   End: