| 1 |
seanius |
175 |
[[!meta title="DEP-10: parallelized ('rolling') release management"]] |
| 2 |
|
|
|
| 3 |
|
|
Title: parallelized ('rolling') release management" |
| 4 |
|
|
DEP: 10 |
| 5 |
|
|
State: DRAFT |
| 6 |
|
|
Date: 2011-04-30 |
| 7 |
|
|
Drivers: Sean Finney <seanius@debian.org>, |
| 8 |
|
|
Raphaƫl Hertzog <hertzog@debian.org> |
| 9 |
|
|
URL: http://dep.debian.net/deps/dep10 |
| 10 |
|
|
License: GPL |
| 11 |
|
|
Abstract: |
| 12 |
|
|
Proposal for changes to release management methodology and infrastructure |
| 13 |
|
|
to remove the requirement that the testing suite must freeze during |
| 14 |
|
|
the release process. |
| 15 |
|
|
|
| 16 |
seanius |
181 |
[[!toc ]] |
| 17 |
|
|
|
| 18 |
seanius |
178 |
<a name="introduction"> |
| 19 |
seanius |
176 |
# Introduction / Problem scope |
| 20 |
|
|
|
| 21 |
|
|
Currently, as the project nears a new stable release, a freeze is instituted |
| 22 |
|
|
on the testing suite. The freeze is put in place to allow the release team |
| 23 |
|
|
to focus on resolving the remaining Release Critical (RC) bugs for the next |
| 24 |
|
|
stable release, and at the same time to prevent regressions from new uploads. |
| 25 |
|
|
Typically the freeze begins as an advisory "soft freeze", which over |
| 26 |
|
|
time increases in strictness and levels of enforcement. |
| 27 |
|
|
|
| 28 |
|
|
Unsuprisingly, as the strictness of the freeze increases, there is an |
| 29 |
|
|
inversely proportional decrease in other non-release targeted maintainer |
| 30 |
|
|
activity. Since unstable still is the preferred route for packages to |
| 31 |
|
|
reach the new release during this period, maintainers are highly discouraged |
| 32 |
|
|
and in some cases prevented from doing non-release targeted activities |
| 33 |
|
|
in unstable. |
| 34 |
|
|
|
| 35 |
|
|
The reduction of such non-release activity is viewed as problematic in |
| 36 |
|
|
this DEP, for some inter-related reasons: |
| 37 |
|
|
|
| 38 |
|
|
* New features and innovations are put on hold, or at least not commonly |
| 39 |
|
|
available, until after the release is made. |
| 40 |
|
|
* Overall Maintainer activity decreases as freezes persist. |
| 41 |
|
|
* Potential userbase is lost (missing feature X, switch distro). |
| 42 |
|
|
* The volatility of testing/unstable increases (and thus quality decreases) |
| 43 |
|
|
with a deluge of new uploads after the freeze is lifted. |
| 44 |
|
|
|
| 45 |
|
|
As Debian is well known for taking a "release when it's ready" approach, the |
| 46 |
|
|
freeze periods are generally known to last considerable amounts of time. |
| 47 |
|
|
Consider the last three freezes: |
| 48 |
|
|
|
| 49 |
|
|
* squeeze: 4 months |
| 50 |
|
|
* lenny: 7 months |
| 51 |
|
|
* squeeze: 6 months |
| 52 |
|
|
|
| 53 |
|
|
This means, in rough terms, that the when testing thaws, that the Debian |
| 54 |
|
|
project may be starting from a state half a year behind comparable |
| 55 |
|
|
distributions. |
| 56 |
|
|
|
| 57 |
|
|
While the resulting stable releases are well known for their high level |
| 58 |
|
|
of quality and stability, it is at a considerable cost. The project is, |
| 59 |
|
|
in essence, starting from a 6 month standstill, will be similarly |
| 60 |
|
|
out of date with comparable distributions. Furthermore, both the standstill |
| 61 |
|
|
as well as the subsequent rush of uploads will introduce further delays |
| 62 |
|
|
to the next release, as release goals will likely be set relative to |
| 63 |
|
|
the starting point. |
| 64 |
|
|
|
| 65 |
seanius |
178 |
# Past and present release methods |
| 66 |
|
|
|
| 67 |
|
|
## Frozen (< 2000) |
| 68 |
|
|
|
| 69 |
|
|
Before the introduction of testing, Debian used a simple release process |
| 70 |
|
|
where the unstable suite was snapshotted into a `frozen` suite. This suite |
| 71 |
|
|
would then be used exclusively for preparing the next stable release, with |
| 72 |
|
|
unstable continuing in parallel. |
| 73 |
|
|
|
| 74 |
|
|
Before freeze Freeze Release |
| 75 |
|
|
|
| 76 |
|
|
[unstable/sid]-------------------------------------------------------------- |
| 77 |
|
|
\ |
| 78 |
|
|
[frozen].-.-.-.-.[stable/R_N].-.-.-.-.-.-.-.-.-.-.-.-.-. |
| 79 |
|
|
|
| 80 |
|
|
[stable/R_N-1].-.-.-.-.-.-.-.-.-.-.-.[oldstable/R_N-1].-.-.-(EOL) |
| 81 |
|
|
|
| 82 |
|
|
--------: Normal activity. Standard rules for uploads and migrations. |
| 83 |
|
|
.-.-.-.-: Release targeted activity. Freezes and limited uploads. |
| 84 |
|
|
\ \ \ \ : Package migration activity. |
| 85 |
|
|
|
| 86 |
seanius |
179 |
### Use cases with `frozen` |
| 87 |
seanius |
178 |
|
| 88 |
seanius |
179 |
Users were divided into two sets, those using `unstable` and those using |
| 89 |
|
|
`stable`. Very few users would use both, as the two lines of development |
| 90 |
|
|
would quickly diverge from each other. |
| 91 |
seanius |
178 |
|
| 92 |
seanius |
179 |
### Benefits with `frozen` |
| 93 |
|
|
|
| 94 |
seanius |
178 |
* Work in unstable continued, unaffected by the freeze. |
| 95 |
|
|
|
| 96 |
seanius |
179 |
### Problems with `frozen` |
| 97 |
seanius |
178 |
|
| 98 |
|
|
* Release work started from a buggy suite. |
| 99 |
|
|
* Duplicated effort in unstable/frozen. |
| 100 |
|
|
* RM had more work/responsibilities, and was prone to burn-out. |
| 101 |
|
|
|
| 102 |
|
|
## Testing (2000-Present) |
| 103 |
|
|
|
| 104 |
|
|
The testing suite was introduced in Debian between the release of potato |
| 105 |
|
|
and woody, in the fall of 2000[[1]]. The goal was to provide |
| 106 |
|
|
a suite that was in a better state for release preparation, by having |
| 107 |
|
|
both automated and manual tools to keep down the level of bugs and |
| 108 |
|
|
general volatility. |
| 109 |
|
|
|
| 110 |
|
|
As a pleasant and convenient side-effect, the new suite also provided |
| 111 |
|
|
a "slightly less buggy unstable" for developers and end-users, who wanted |
| 112 |
|
|
newer software/features not available in stable, but wanted some level |
| 113 |
|
|
of protection to the relatively unpredictable nature of unstable. |
| 114 |
|
|
|
| 115 |
|
|
Before release Freeze Release |
| 116 |
|
|
|
| 117 |
|
|
[unstable/sid]----------.--.--.--.-.-.-.-.-.-.------------------------------ |
| 118 |
|
|
\ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ |
| 119 |
|
|
[testing/R_N]----------.-.-.-.-.-.-.-.-.-.-.-.[testing/R_N+1]--------------- |
| 120 |
|
|
/ / \ |
| 121 |
|
|
/ / [stable/R_N].-.-.-.-.-.-.-.-.-.-.- |
| 122 |
|
|
/ / / / / |
| 123 |
|
|
[R_N p-u].-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.- |
| 124 |
|
|
|
| 125 |
|
|
[stable/R_N-1].-.-.-.-.-.-.-.-.-.-.-.-.-.-[oldstable/R_N-1].-.-.-(EOL) |
| 126 |
|
|
|
| 127 |
|
|
--------: Normal activity. Standard rules for uploads and migrations. |
| 128 |
|
|
.-.-.-.-: Release targeted activity. Freezes and limited uploads. |
| 129 |
|
|
\ \ \ \ : Package migration activity. Spacing of marks is a rough |
| 130 |
|
|
indication of frequency. |
| 131 |
|
|
|
| 132 |
|
|
During the freeze, the testing suite becomes entirely dedicated to the |
| 133 |
|
|
release work. In practice, this also means that unstable is also to |
| 134 |
|
|
some extent frozen from non-release activity for many packages, since |
| 135 |
|
|
it still serves as the main route to testing for new uploads. |
| 136 |
|
|
|
| 137 |
seanius |
179 |
### Use cases with `testing` |
| 138 |
seanius |
178 |
|
| 139 |
seanius |
179 |
The introduction of the new suite also introduced a type of continuum |
| 140 |
|
|
in which users now have more flexibility in selecting what to have installed. |
| 141 |
|
|
|
| 142 |
|
|
* `unstable`: only the "bleeding edge" of Debian updates. |
| 143 |
|
|
* `testing`: only the "leading edge" of Debian updates. |
| 144 |
|
|
* `testing/unstable`: A hybrid solution of the two previous use cases, |
| 145 |
|
|
allowing for a reasonable balance between them. |
| 146 |
|
|
* `stable`: The latest stable release. |
| 147 |
|
|
* `stable/testing`: the stable release with perhaps some newer packages |
| 148 |
|
|
installed, typically also making use of "APT pins" to prevent unwanted |
| 149 |
|
|
upgrades. Less common now with the inclusion of backports and volatile |
| 150 |
|
|
as official Debian services. |
| 151 |
|
|
|
| 152 |
seanius |
180 |
### Benefits with `testing` |
| 153 |
seanius |
179 |
|
| 154 |
seanius |
178 |
* Release branch has far fewer bugs at the start of the release process. |
| 155 |
|
|
* (Ideally) shorter release process with fewer problems. |
| 156 |
|
|
* Large user-base for testing stable-targeted fixes. |
| 157 |
|
|
|
| 158 |
seanius |
179 |
### Problems with `testing` |
| 159 |
seanius |
178 |
|
| 160 |
|
|
(See [Introduction](#introduction)) |
| 161 |
|
|
|
| 162 |
|
|
[1]: http://lists.debian.org/debian-devel/2000/08/msg00906.html |