The Gnu Writing Movement (GWM) was started from scratch with one goal: improving and reorganising the GNU documentation, using the Linux Documentation Project (LDP) experience as a background.
Improving does not mean competing - against the LDP, KDP, GDP or any other documentation project. The GWM technical innovations will be shared with the other projects, under the GNU Public License (GPL).
Free Software Documentation, which should be called "Libre Documentation," is documentation one can copy or enhance, just like free software, as long as those freedoms are not restricted.
We say "libre" instead of "free" because libre means free as in free speech, and is not about financial prices but liberty and freedom.
Besides increasing awareness on Libre Documentation and promoting Libre Documentation licenses, such as the Gnu Free Documentation License (GNU FDL or GFDL) the following is considered as the base of the GWM tasks:
There is already a lot of documentation for free software available on the Internet in various places. Most of it has been released under a free license by the existing documentation projects, or on individual websites.
Before considering an effort to improve documentation, careful checking of the existing documentation is required to avoid duplicating work whenever possible. If different documents exist for the same topic, they may be recommanded as alternatives documents or suggested readings.
A unique ID for each document is the base of the whole system, especially when documents from outside the GWM are considered.
Assigning an ID to each document means it would be tracable, and linked to additional information on a database.
This traceability will make possible to "follow" that document when new releases are made, when translations are submitted and when alternatives are found. It will as well ease indexing and retrieving, by both the search engine and the documentation browser on the user computer.
For example, a search at Google.com on "bash documentation" returns the official info page, online courses, the official Bash FAQ, and the LDP Advanced Bash-Scripting Guide in the results.
An ID is given to each document, and provide more information of the document. We can also link to the other IDs sharing the topic.
1-1-en-texinfo title "bash info page"
1-1-en-texinfo URL "http://www.gnu.org/manual/bash/texi/bashref.texi.tar.gz"
1-1-en-texinfo keywords "bash, shell, command line, command prompt"
1-1-en-texinfo userrating "250"
1-1-en-texinfo alternatives 2,3,4
2-1-en-html title "free-ed bash courses"
2-1-en-html URL "http://www.free-ed.net/fr03/lfc/course%20030109_03/lesson10.htm"
2-1-en-html keywords "bash, shell, tutorial, course, beginner"
2-1-en-html userrating "210"
2-1-en-html alternatives 1,3,4
3-1-en-txt title "the official bash faq"
3-1-en-txt URL "http://dougbarton.net/Bash/Bash-FAQ.txt"
3-1-en-txt keywords "bash, faq, questions"
3-1-en-txt userrating "255"
3-1-en-txt alternatives 1,2,4
4-1-en-pdf title "Advanced Bash-Scripting Guide"
4-1-en-pdf URL "http://www.linuxdoc.org/LDP/abs/abs-guide.pdf"
4-1-en-pdf keywords "bash, shell, guide, reference"
4-1-en-pdf userrating "200"
4-1-en-pdf alternatives 2,3,4
4-1-en-html title "Advanced Bash-Scripting Guide"
4-1-en-html URL "http://www.linuxdoc.org/LDP/abs/html/index.html"
4-1-en-html keywords "bash, shell, guide, reference"
4-1-en-html userrating "200"
4-1-en-html alternatives 2,3,4
For compatibility reasons, using the Dublin Core metadata tags and format are preferred, It can be completed if needed by our own tags such as the known alternatives. The list of fields the GWM will support has yet to be selected: title/url/keywords/ratings/alternatives example may be completed by other fields such as date, copyright, author name, license, etc.
The ID should be put inside the documentation, using a special tag to refer to it, which should not be a problem for GWM documents. However authors of non-GNU documentation may be reluctant to add an indexation ID.
Some fields must be included within the ID itself, such as the original resource identifier (3 for Bash FAQ), version number (1), language (en) and document type (txt), to ease documentation browser work for documentation updating.
For example, the next version of the Bash FAQ in spanish will be 3-2-sp-txt. The precise URL, keyword, license or date are irrelevant when it comes to version checking! Any user reading the official Bash FAQ (3-1-en-txt) should have an easy way to check for new versions. The documentation reader or any kind of software must take care of that process with very little information (mostly what the user currently reads). It must then offer the possibility to download and read the new (3-2-en-txt) version. For example, if the locale has been set to ES, (3-1-sp-txt) should be offered as an alternative. And when 3-2-sp-txt is available (the latest version in the user preferred language), it should be put at the top of the list. Lower on the list, documents 1-1-en-texinfo and 2-1-en-html (with the highest ratings after the FAQ) should be suggested as alternative readings.
Knowing the exact date of the new release, the precise URL or the license (whether it change between each version) is interesting, but only when a document suited for the user has been found. This additional metadata information may then be accessed querying the database or any other system associating the ID to the complete document information.
As previously stated, the ID should be included within the document using a special tag meaning "here is the document ID". Other methods, such as using the ID as the filename, or putting the metadata in a separate file of the same name but a different extension may be perfect for websites (after which Dublin Core metadata was mostly designed), but a poor solution for documentation which can be printed or stored on various repositories as well.
For example, if the Bash FAQ in french is printed, 3-1-fr-txt will be somewhere at the beginning. With this ID, additional information, new versions, different languages (etc.) can easily be found when checking on the GWM website.
Suppose the French user who printed the FAQ has a question on bash, or a suggestion for the author. Sending the question to a Usenet newsgroup and the suggestion to the author email are the current best practices.
However, the author can not respond to every single message, and many people are not looking for the answer to their question on Usenet, assuming they are the very first one to experience that very problem.
Two specific fields containing forums URL (document specific, and topic specific) should therefore be included in the metadata of each document. The URL could be a Usenet newsgroup, an online web forum or anything else.
Document specific forums would be perfectly suited for suggestions to the author, corrections or additions, while topic specific forums would be used to request help. Forums could be identical for a suite of documents.
Documents 1-1-en-texinfo 2-1-en-html 3-1-en-txt and 4-1-en-pdf could for example share a "bash" forum. Multiple forums could also be listed with broader subjects like "configuring email".
Should the unique ID be applied to every document, both the search engine and the documentation reader could take advantage of it to provide a high quality indexation of online resources.
There are two risks: abuse (such as keyword abuse for websites to get a higher place on the search engines and poor completion of metadata fields), and duplication.
A possible solution would be reserving ID assignation and metadata filling to documentation projects (such as the GWM, the LDP, etc.). Each documentation project would maintain its own base of ID for the documents maintained within the documentation project, and share his base with other projects at regular intervals. This would make it possible for each documentation project to display in its search results documents coming from other documentation projects.
Any organisation can provide its own ID as well, and non-affiliated authors could request an ID for their document at the documentation project of their choice.
Using the ID assignation source as yet another ID direct field would make synchronisation and searching easy.
For example, consider the Bash FAQ, maintained by an independent author, requests an ID at the GWM and is assigned 3-1-en-txt. But the volunteer based ID assigning effort at the LDP gave the document ID 56-2-en-txt. The document would have 2 IDs: GWM-3-1-en-txt and LDP-56-2-en-txt. When the KDP, the GDP, the GWM and the LDP ID servers synchronise, these two new entries would be mirrored on each system. Using simple robots to check for similarities or duplicates ('uniq' for URL, title, etc.) would remove this risk. In this case, the LDP may decide to remove its duplicate entry.
Now imagine the SDP (Spam Documentation Project) creates a "Hot Erotic Bash Win Millions While Working At Home FAQ" of the document, pointing to some non-documentation URL and abusing metadata keywords (porn, porn, bash, bash, bash, hot, porn, bash) to get high indexing. It self assigns SDP-1-1-en-txt ID. And the CDG (Closed-Source Documentation Group) creates a non-free document, "The Best Bash FAQ Ever," to which it assings CDG-55-0-en-txt.
Should there entries be included the KDP/GDP/GWM/LDP synchronisation ring, removing them would be easier. The GNU Writing Movement could refuse to display closed source documents, which CDG makes, therefore barring any CDG ID from its search engine list of results. And every documentation project could refuse to transport SDP documents, which metadata is known to be falsified.
Suppose a Canadian woman may want to write brand new documentation in English, while a Norvegian man may want to provide a Nynork version of a manual.
These are different ways of contributing but in the end they enhance GNU documentation. Currently documentation is handled within each project, where it is tied to a program version. In some cases, the documentation is outdated. And documentation volunteers don't know where to go - these people who like the free software ideas would like to contribute but since they don't know how to code will not be pointed to the project team and its outdated documentation.
Therefore all documentation should be revived by the GWM and put on the GWM "todo list", which would act as an entry point for volunteers as well. For example, WXYZ package reports its documentation is badly outdated, TUVW needs translation in turkish and ABCD is fine. The volunteers could receive this information by email on the GWM lists and therefore assign themselves depending on their knowledge, their avalability and the documentation needs.
At each conference or meeting, people who want to help but can't code could be given the GWM address, where the documentation needs are going to be listed.
Documents which are hard to understand, technically inadequate or outdated will be handled by a specific unit of the GWM - a "Quality Check" peer reviewing group. It can update documentation status on the GWM todo list. For example, ABCD documentation is considered not to be understandable by beginners after peer reviewing, while ABCD team thought the documentation was fine.
Writing documentation consumes more time than maintaining documentation. This task should not be assigned to 'rookie volunteers' unless they specifically request it, for it is a time consuming and difficult task.
Yet if the IJKL package, which newly joined the GNU project, needs documentation it should be able to ask the GWM. If no volunteer could be found within the GWM, a member of IJKL team could be taught which tools he should use, which draft of documentation he could use as a starting basis, etc.
A monitoring of the writing by the GWM authors and "Quality Check" team could mean a new document which doesn't need to be enhanced afterwards.
The GWM will offer authors and translators help in writing, proof reading and converting documentation. Sub-groups will also be formed to translate the documentation in as many languages as possible, and finally produce packages in different formats (including Texinfo) which could go in many different places, such as a website (html), a ftp server (pdf or ps) and the synchronisation queue of the documentation reader.
Any finished documentation will be passed to the package team, which could also request a current draft, to be included with each package.
The documentation would not be tied to the software development - a volunteer for each package would be the link between the software package and us. This role would be very important in the documentation authoring process, giving the global orientation of the documentation, synchronising it with the software version or suggesting new chapters.
The volunteer "link" would not be an author unless he or she wanted but, but would read the GWM mailing lists and help the authors in change of this document.
The authors would take advantage of automatation: for example, they could send new versions of documents by CVS or by email, signed the gpg key of the author. The document would be checked, then the document it be put into the GWM database from which a cron file would generate the needed outputs, mail announcements of the new document, put the news in the mainpage of the GWM, notify the translation groups and send them a diff, update the metadata to increase version number, etc.
Authors may be interested in having new versions of their work online by the next day - only new submissions would require manual processing (for the initial metadata, and document approval).
Joe User is teaching Bash for a living in Québec. He needs the Bash FAQ and the Advanced Scripting guide in French and in English. He best likes the GDP, because there is a local mirror in Montréal.
His documentation reading application has GWM-4-1-en-pdf, GWM-4-1-fr-pdf, LDP-56-2-en-txt in the favorites. His documentation reading application is synchronising to the Montréal GDP mirror, which itselfs synchronises to the the LDP and GWM server.
If a new version of the FAQ in English is released, the GDP server will receive GWM-4-2-en-pdf. The next time Joe's documentation reader synchronises, it should automatically retrieve this document, and warn him a new version is now available. It should also tell him LDP-56-2-fr-txt is available as well. When the LDP-56-2-fr-txt/GWM-3-2-txt duplicate is removed from the GDP, it should automatically use GWM-3-2-txt.
Joe is quite happy with the new version of the FAQ in English, but finds an obvious typo. Mailing the bug report with GNOME Documentation Reader is easy - it's automatically sent to the email associated with the document, in which case the documentation reader opens a mail client to mailto:firstname.lastname@example.org where bash-faq maintainers will read the message. The list archives being available on the web, anybody going to the document public forum will be able to see the bug report. An author can correct the typo in the document, write a new version, gpg sign it and submit it to the GWM where it will be available within hours. Joe will synchronise to this document automatically.
When Joe goes to the GDP website to search for "bash courses", GWM-2-1-en-html is presented (the online courses). He could have requested "suggested readings" to his Documentation Browser to find this document as well.
Anyway, he really likes what he sees - so much that he wants to write a French version. For that he simply goes to gwm.gnu.org, register online as a new author and sends his document, which will be assigned within 4 days the GWM-2-1-fr-html ID by the GWM team. The ID will be propagated to collaborating documentation projects. When GWM-2-2-en-html is out, this new version will automatically be announced to Joe since he is the French translation maintainer, so that he can update it. He will be provided a diff to ease his translation work. GWM-2-2-fr-html will be automatically put online within hours since Joe signed the new version with his gpg key and sent it to the GWM.
When GWM-2-2-en-html was just out, another message was sent to the online course forum, where a member of the LDP read it. He came there to reply with help for a beginner question in the first place. Now he thinks he should help with the new version of the document, or with the info page which is not as good as the course introduction is. He knows that because he had set up his documentation reader to synchronise with the whole GWM and LDP collection of document, for offline reading.