gpgme/lang/python/doc/src/maintenance-mode.org

136 lines
6.1 KiB
Org Mode
Raw Permalink Normal View History

# -*- mode: org -*-
#+TITLE: Maintenance Mode
#+AUTHOR: Ben McGinnes
#+LATEX_COMPILER: xelatex
#+LATEX_CLASS: article
#+LATEX_CLASS_OPTIONS: [12pt]
#+LATEX_HEADER: \usepackage{xltxtra}
#+LATEX_HEADER: \usepackage[margin=1in]{geometry}
#+LATEX_HEADER: \setmainfont[Ligatures={Common}]{Times New Roman}
#+LATEX_HEADER: \author{Ben McGinnes <ben@gnupg.org>}
* Maintenance Mode from 2019
:PROPERTIES:
:CUSTOM_ID: maintenance-mode
:END:
| Version: | 0.0.1 |
| GPGME Version: | 1.13.0 |
| Author: | Ben McGinnes <ben@gnupg.org> |
| Author GPG Key: | DB4724E6FA4286C92B4E55C4321E4E2373590E5D |
| Language: | Australian English, British English |
| xml:lang: | en-AU, en-GB, en |
From the beginning of 2019 the Python bindings to GPGME will enter
maintenance mode, meaning that new features will not be added and only
bug fixes and security fixes will be made. This also means that
documentation beyond that existing at the end of 2018 will not be
developed further except to correct errors.
Though use of these bindings appears to have been quite well received,
there has been no indication of what demand there is, if any for
either financial backing of the current Python bindings development or
support contracts with g10code GmbH citing the necessity of including
the bindings.
** Maintainer from 2019 onward
:PROPERTIES:
:CUSTOM_ID: maintenance-mode-bm
:END:
How does this affect the position of GnuPG Python Bindings Maintainer?
Well, I will remain as maintainer of the bindings; but without funding
for that position, the amount of time I will be able to dedicate
solely to this task will be limited and reduced to volunteered time.
As with all volunteered time and effort in free software projects,
this will be subject to numerous external imperatives.
** Using the Python Bindings from 2019 and beyond
:PROPERTIES:
:CUSTOM_ID: maintenance-mode-blade-runner
:END:
For most, if not all, Python developers using these bindings; they
will continue to “just work” the same as they always have. Expansions
of GPGME itself are usually handled by SWIG with the existing code and
thus bindings are generated properly when the bindings are installed
alongside GPGME and when the latter is built from source.
In the rare circumstances where that is not enough to address some new
addition to GPGME, then that is a bug and thus subject to the
maintenance mode provisions (i.e. it will be fixed following a bug
report being raised and your humble author will need to remember where
the timesheet template was filed, depending on how many years off such
an event is).
All the GPGME functionality will continue to be accessible via the
lower level, dynamically generated methods which match the GPGME C
documentation. While the more intuitively Pythonic higher level layer
already covers the vast majority of functionality people require with
key generation, signatures, certifications (key signing), encryption,
decryption, verification, validation, trust levels and so on.
Any wanted features lacking in the Python bindings are usually lacking
because they are missing from GPGME itself (e.g. revoking keys via the
API) and in such cases they are usually deliberately excluded. More
discussion of these issues can be found in the archives of the
[[https://lists.gnupg.org/mailman/listinfo/gnupg-devel][gnupg-devel mailing list]].
Any features existing in the dynamically generated layer for which
people want a specific, higher level function included to make it more
Pythonic (e.g. to avoid needing to learn or memorise cryptographic
mode values or GnuPG status code numbers), would be a feature request
and /not/ a bug.
It is still worthwhile requesting it, but the addition of such a
feature would not be guaranteed and provided on a purely volunteer
basis. Expediting such a request would require funding that request.
Those with a commercial interest in expediting such a feature request
already know how to [[https://gnupg.org/cgi-bin/procdonate.cgi?mode=preset][expedite it]] (use the message field to state what
feature is being requested).
** Documentation formats
:PROPERTIES:
:CUSTOM_ID: docs
:END:
The documentation has been written in Org mode for GNU Emacs, with
both Texinfo and reStructuredText formats generated from that. The
Texinfo files are intended for use with the rest of the GnuPG
documentation; while the reStructuredText files are intended for use
with Docutils and Sphinx, as with other Python projects.
*** Cautionary Notes regarding Sphinx and EPUB
:PROPERTIES:
:CUSTOM_ID: sphinx-made-epubs-suck
:END:
Though Python's Docutils in conjunction with Sphinx is capable of
generating some very useful HTML sites, as proven by [[https://readthedocs.org/][Read the Docs]] and
the [[https://docs.python.org/][Python documentation]], there are a number of output formats it does
not handle well. At the top of the list of things it manages to break
so atrociously as to be embarassing is the [[http://idpf.org/epub][EPUB 3]] format.
The automatically generated EPUB of the CPython documentation always
contains hundreds of validation errors and even the modest amount of
documentation here [[https://files.au.adversary.org.s3.amazonaws.com/crypto/gpgme-python/rst/epub/GPGMEPythonBindings.epub][produced a file]] with approximately thirty
validation errors. As the volume of documentation content increases,
so does the induced errors. Whereas Texinfo doesn't produce EPUB
output at all, nor does Org-mode.
Should there ever be genuine demand for this format, lodge a [[https://dev.gnupg.org/maniphest/task/edit/form/4/][feature
request]] case marked for [[https://dev.gnupg.org/p/BenM/][my]] attention. The means of generating such
files flawlessly is already available, but is not yet part of the
GnuPG build system. Nor is it integrated with a means of converting
Org mode input files to the relevant base format automatically, as can
already be done when converting Org to reStructuredText or Org to
Texinfo. As a certain amount of work would be required to get it
done, there would need to be clear demand for that work to be done.