summaryrefslogtreecommitdiff
path: root/contributing/web_pages/news.mdwn
blob: 4c09e1abf330c85fe8355e9cc7f55ef2e40b56fa (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
[[!meta copyright="Copyright © 2009 Free Software Foundation, Inc."]]

[[!meta license="""[[!toggle id="license" text="GFDL 1.2+"]][[!toggleable
id="license" text="Permission is granted to copy, distribute and/or modify this
document under the terms of the GNU Free Documentation License, Version 1.2 or
any later version published by the Free Software Foundation; with no Invariant
Sections, no Front-Cover Texts, and no Back-Cover Texts.  A copy of the license
is included in the section entitled [[GNU Free Documentation
License|/fdl]]."]]"""]]

[[!meta title="""How to prepare and publish "a month of the Hurd" for the last
month"""]]

We prepare *a month of the Hurd* in a public git branch (`master-news_next`),
and merge that one into `master` once we want to publish the news.  The idea is
to record to-be-published changes on that branch at they time they arise, and
then publish them en bloc at the end of the month.

As this is done on a separate branch, there are two options: you can have
separate repositories (*clones*, or *checkouts*; what you get from `git clone`)
for the `master` and `master-news_next` branches, or you can deal with both in
the same repository.  Having separate repositories you don't have to remember
which branch you're on, and don't have to switch between branches before
beginning to edit files, and it doesn't matter -- as no switching between
branches is needed -- if you have uncomitted changes to some files.  On the
other hand, you may want to keep it all in one repository, to save disk space,
and for shuffling different branches' commits being (a bit) easier.

For practical work that means to use the following commands: 

  * create a local branch for `master-news_next`

        $ git fetch
        $ git checkout -b master-news_next origin/master-news_next

      * if you already have created a local branch `master-news_next`: update to
        the latest version of it

            $ git checkout master-news_next
            $ git pull --rebase

  * edit

    Always do check which branch you're on (asterisk in the first column of
    `git branch`'s output), and only then begin to do your changes and commit
    them.

    We should not update news items that have already been published (that is,
    on <http://www.gnu.org/software/hurd/news.html>; no problem for the
    [[staging area|web_pages#staging_area]]), as the system will always also
    update the RSS feeds, etc., causing the old news item to reappear on feeds,
    which is a bit of a nuisance.

      * at the beginning of a month: create a new news entry

            $ cp news/2009-07-31.mdwn news/YYYY-MM-DD.mdwn
            $ # edit the new file
            $ git add news/YYYY-MM-DD.mdwn
            $ git commit -m "Begun the news entry for YYYY-MM-DD."

        That is, use the *2009-07-31* news snippet as a template for the new
        one.

  * check the outgoing changes

        $ git log --reverse -p -C origin/master-news_next..master-news_next

  * push the changes

        $ git push origin master-news_next

      * if push fails, pull and rebase the local changes on top of the remote
        changes

            $ git pull --rebase

        This will only happen if someone else has been installing commits into
        `origin`'s `master-news_next` branch since the last time you
        synchronized to it.

        Note that this might cause some conflicts to arise -- if the remote
        repository contains commits that conflict with any that you've been
        recording in your local repository.  For this reason, you might want to
        already do this *rebase* before beginning you local edits, simply to
        shorten the time frame in which such conflicts can arise.
        (Theoretically, in the very rare case of very much concurrent editing
        going on, you'd have to repeat this again (and again...) before
        succeeding to push your changes.)

  * at the end of the month: prepare for publishing the news

    Edit the news entry's *meta date* and *meta updated* time values.  For
    stability, we set these two explicitly instead of relying on ikiwiki's
    method of using the files' creation and modification dates.

  * ... and publish

        $ git checkout master
        $ git pull --rebase
        $ git merge master-news_next
        $ git push origin master

    That means, for publishing we merge `master-news_next` into `master`.
    After that merge, work for the next month's news item can continue on
    `master-news_next`.

And since we're working in a branch anyway: Please commit often, push often and pull often! Few things would be nastier than having to do heavy merging on a enws release date with little time at hand.