The purpose of this analysis is to estimate the usefulness of the ACK/NACK system as used in OpenVPN project's development. This study goes back two years, and the data is obtained from Git logs on a Git clone taken on 5th April 2012:
$ git log --since="2 years ago"
This data is cross-referenced with the emails stored in the openvpn-devel mailing list archive. As reviewing all patches and ACKs is not possible, a random sample is taken.
The hypothesis is that the ACK/NACK process is most suitable for reviewing individual patches or small patchsets sent to the openvpn-devel mailing list by developers not intimately familiar with OpenVPN. A less bureaucratic process might be more suitable for core developers, especially when less security-sensitive parts are being modified.
Current ACK/NACK code review system
Most patches have required an ACK before going into Git "master". There are a only few exceptions:
- Patches from James Yonan's Subversion repository are merged without review
- Anything else?
The original ACK system required an ACK from one developer for the code to enter the Git repository. This process was later (add date?) split into two logical parts:
- Feature-ACK: this should be the primary consideration when reviewing patches. Merging useless patches makes no sense whatsoever.
- Code-ACK: if the feature makes sense, the next step would be to review that the code itself makes sense (e.g. has no obvious errors and follows established conventions)
Still later (add date?) a move towards lazy consensus was taken. So, as it stands now (Mar 2012), code can go to the Git repository without an ACK, but with a considerable delay.
The formal ACK/NACK system was imposed to increase OpenVPN code quality by decreasing the chances of bad code getting into the tree. The mandatory ACK forces peer review. In software engineering jargon, it fulfills the roles of
The review process obviously helps most when less-knowledgeable people send in patches which often have obvious errors and omissions.
There are limits to what the ACK process can do. For example:
- Runtime errors may be very difficult to notice
- Seeing the "big picture" may be difficult, especially with big patchsets
- Whether a particular problems is noticed depends highly on the reviewer's skills
The ACK/NACK process has it's cons:
- Requiring an ACK slows down flow of code from contributors to the "master" development tree. This problem gets worse as the patchset size increases.
- Due to the above, it takes longer to get the code into wide circulation. This means bugs are reported later than if the code had been merged as soon as it's ready.
This raw data is based on Git clone done on 5th Apr 2012:
- Commit count: 635
- ACK number: 458
- Average ACK count: .72
ACKs by person
- Adriaan de Jong 20
- Alon Bar-Lev 3
- Davide Guerri 1
- David Sommerseth 155
- Eric F Crist 1
- Gert Doering 107
- Gilles Espinasse 1
- Heiko Hund 3
- James Yonan 116
- Jan Just Keijser 2
- Kazuyoshi Aizawa 2
- krzee 3
Effects of the ACK system on patches
The following Git command was used to generate a list of all patches in Trac format:
$ git log --date=short --since="2 years ago" --pretty=format:"||%ci||%ad||%h||%s||-||%an||-||-||-||"
A random sample of 64 patches (out of 634) was then taken using shuf -n 64. A few notes:
- Delay is calculated as "day when applied to git" - "day when sent to the ml". For example, 26th Mar - 24th Mar = 2 days. When calculating large (i.e. 1+ month) delays each month is counted as 30 days.
- Patchset refers to a major patchset; in practice it's either Buildsystem or PolarSSL.
- Author date is the date the patch was introduced (e.g. sent to the mailing list)
- Commit date is the date when the patch was commited to current Git master repository. Note that changes to Git repository's structure (e.g. "allmerged -> master") seem to have messed up Git's commit dates. This means that some patches introduced before 2010-07-27 seem to have been "floating around" for months, whereas in reality they were merged to Git fairly quickly. Attempts to find out real dates often failed, so the faulty numbers were left as-is.
- For many patches it was impossible to locate the ACK - see "?" in "Delay (after ACK)" column. These may represent ACKs given in the IRC outside IRC meetings. ACKs given in the official IRC meetings should pop up when searching the mailing list archives.
- In some cases it was impossible to determine where a patch was originally sent to. In these cases figuring out whether it was modified in the process was also impossible.
|Commit date||Author date||Identifier||Patch name||Patchset||Author||Modified||Delay (total)||Delay (after ACK)|
|2012-03-30||2012-03-28||3144411||Enable pedantic in windows compilation||Buildsystem||Alon Bar-Lev||No||2||0|
|2012-03-26||2012-03-24||405f338||build: windows: set vendor to openvpn project + cleanups||Buildsystem||Alon Bar-Lev||No||2||0|
|2012-03-22||2012-02-29||fbae7d2||build: plugins: properly use CC, CFLAGS and LDFLAGS||Buildsystem||Alon Bar-Lev||No||23||7|
|2012-03-22||2012-02-29||553d95d||cleanup: memcmp.c: remove unused source||Buildsystem||Alon Bar-Lev||No||23||16|
|2012-03-22||2012-02-29||51bd56f||build: autotools: first pass of trivial autotools changes||Buildsystem||Alon Bar-Lev||No||23||7|
|2012-02-20||2012-02-18||4ebc587||define access mode flag X_OK as 0 on Windows||-||Heiko Hund||Unknown||2||-|
|2012-03-30||2012-02-14||8e5613c||Migrated x509_get_sha1_hash to use the garbage collector||-||Adriaan de Jong||Yes||45||3|
|2012-01-23||2012-01-22||62c613d||Platform cleanup for FreeBSD||-||Gert Doering||No||1||0|
|2011-12-14||2011-11-03||8407991||Fixed client issues with DHCP Router option extraction/deletion when using layer 2 with DHCP proxy:||-||James Yonan||No||41||-|
|2011-11-21||2011-10-31||7ac7170||Reordered functions to ensure warning-free Windows build||-||Adriaan de Jong||No||21||11|
|2011-11-21||2011-10-31||54628d1||Moved prng_uninit out of crypto_uninit_lib||-||Adriaan de Jong||No||21||?|
|2011-11-21||2011-10-31||1d90851||Moved from strsep to strtok, for Windows compatibility||-||Adriaan de Jong||No||21||?|
|2011-10-22||2011-10-16||eaacf8d||Moved to PolarSSL 1.0.0:||-||Adriaan de Jong||No||8||?|
|2011-12-14||2011-10-11||359adbf||Raised D_PID_DEBUG_LOW from level 3 to 4 to reduce replay error verbosity at level 3.||-||James Yonan||No||34||-|
|2011-09-21||2011-09-16||8ca19c0||Platform cleanup for NetBSD||-||Gert Doering||No||5||0|
|2011-08-31||2011-08-31||d90428d||add --mark option to set SO_MARK sockopt||-||Heiko Hund||No||0||0|
|2011-08-24||2011-08-19||2627335||CC_PRINT character class now allows any 8-bit character value >= 32. This is done to allow UTF-8 and restrict the use of control characters in usernames, passwords, common names, etc.||-||James Yonan||No||5||-|
|2012-02-21||2011-08-17||14a382a||remove wrapper code for Windows CryptoAPI function||-||Heiko Hund||No||4||0|
|2011-08-24||2011-08-16||f0257ab||For all accesses to "struct route_list 2 3 AUTHORS build, ... ,version.sh.in rl", check first that rl is non-NULL||-||Gert Doering||Unknown||8||?|
|2011-10-22||2011-07-28||c94eff3||Added back checks for ks->authenticated in verify_user_pass||PolarSSL||Adriaan de Jong||No||86||?|
|2011-08-19||2011-07-25||576dc96||Merge remote branch SVN 2.1 into the git tree||-||David Sommerseth||-||-||-|
|2011-10-22||2011-07-05||a9bf901||Added an extra define to allow building without PKCS#11||PolarSSL||Adriaan de Jong||No||109||?|
|2011-10-22||2011-07-05||557624e||Hardening: periodically reset the PRNG's nonce value||PolarSSL||Adriaan de Jong||No||109||?|
|2011-10-22||2011-07-01||8c96419||Fixed a missing include in ssl_backend.h||PolarSSL||Adriaan de Jong||No||113||?|
|2011-10-22||2011-06-30||3e44ea5||Refactored tls-verify script code||PolarSSL||Adriaan de Jong||Yes||114||?|
|2011-07-01||2011-06-29||d22a379||Made domake-win builds to use easy-rsa/2.0/openssl-1.0.0.cnf||-||Samuli Seppänen||No||2||?|
|2011-10-19||2011-06-27||eab0cf2||Refactored TLS_PRF to new hmac and md primitives||PolarSSL||Adriaan de Jong||No||114||?|
|2011-10-19||2011-06-23||b01cb9e||Refactored crypto initialisation functions||PolarSSL||Adriaan de Jong||No||117||?|
|2011-10-19||2011-06-23||4a5a603||Refactored NTLM DES key generation||PolarSSL||Adriaan de Jong||No||117||?|
|2011-06-20||2011-06-20||54c739e||Revert "Add new openssl.cnf to easy-rsa/Windows"||-||David Sommerseth||No||0||0|
|2011-11-25||2011-06-16||0f2bc0d||Do some file/directory tests before really starting openvpn||-||David Sommerseth||No||1||0|
|2011-08-25||2011-05-26||739fa98||Fix a Visual Studio 2008 build error in options.c||-||Samuli Seppanen||No||90||?|
|2011-05-27||2011-05-26||21fc2ed||Don't define ENABLE_PUSH_PEER_INFO if SSL is not available||-||David Sommerseth||No||1||0|
|2011-04-27||2011-04-27||b70d99f||Fix compile issues when using --enable-small and --disable-ssl/--disable-crypto||-||Gustavo Zacarias||No||0||0|
|2011-04-26||2011-04-24||d549726||Added 'dir' flag to "crl-verify" (see man page for info).||-||James Yonan||No||2||-|
|2011-04-26||2011-04-12||e51935d||For Mac OSX, when DARWIN_USE_IPCONFIG is defined ,.. the ipconfig command sometimes fails if executed immediately after the tun device open.||-||James Yonan||No||14||-|
|2011-04-12||2011-04-12||2a12831||For Mac OSX, when DARWIN_USE_IPCONFIG is defined ... the ipconfig command sometimes fails if executed immediately after the tun device open.||-||James Yonan||-||-||-|
|2011-04-14||2011-04-03||9ed122e||Fixed bug in port-share that could cause port share process to crash with output like this:||-||James Yonan||No||11||-|
|2011-03-25||2011-03-25||dc2ccc8||Clarify --tmp-dir option||-||chantra||No||0||0|
|2011-03-21||2011-03-21||1df945e||Version 2.1.3n||-||James Yonan||-||-||-|
|2011-03-17||2011-03-17||d02a86d||Renamed branch to reflect that it is no longer beta.||-||James Yonan||-||-||-|
|2011-03-21||2011-03-15||58704ea||Updated INSTALL-win32.txt||-||Samuli Seppänen||No||6||3|
|2011-03-25||2011-02-28||a75c7dd||Fixed typo in plugin.h||-||Stefan Hellermann||No||25||?|
|2011-02-27||2011-02-11||a7f0fc3||Added configure.h and version.m4 variable parsing to win/config.py||-||Samuli Seppänen||?||16||?|
|2011-02-27||2011-02-11||4e4aa65||Several modifications to win/make_dist.py to allow building the NSI installer||-||Samuli Seppänen||?||16||?|
|2010-11-18||2010-11-15||7581c8f||Removed functions not being used anywhere||-||David Sommerseth||No||3||?|
|2010-11-18||2010-11-15||33ee747||Fixed potential misinterpretation of boolean logic||-||David Sommerseth||No||3||?|
|2010-11-13||2010-11-13||22178d0||Merge branch 'svn-BETA21' into bugfix2.1||-||David Sommerseth||-||-||-|
|2010-11-12||2010-10-30||f0eac1a||Make "topology subnet" work on Solaris (ifconfig + route metric changes by Kazuyoshi Aizawa, adding of local "connected subnet" route by me)||-||Gert Doering||No||12||?|
|2010-10-21||2010-10-21||59afc4a||Fix problem with special case route targets ('remote_host')||-||Gert Doering||No||0||0|
|2010-11-12||2010-09-30||58f8d94||Add HTTP/1.1 Host header||-||Lars Hupel||No||43||40|
|2010-07-27||2010-07-27||75dfe3d||Added "net stop dnscache" and "net start dnscache" in front of existing --register-dns commands.||-||James Yonan||No||0||0|
The delays shown in these patches are probably faulty (i.e. too big):
|Commit date||Author date||Identifier||Patch name||Patchset||Author||Modified||Delay (total)||Delay (after ACK)|
|2010-10-21||2010-07-16||4c91fc8||Fixes openssl-1.0.0 compilation warning||-||chantra||No||95||?|
|2010-10-21||2010-04-27||7d5e26c||Fix certificate serial number export||-||Davide Brini||No||172||?|
|2010-10-21||2010-04-22||892e64b||Fix missing /bin/bash -> /bin/sh||-||Davide Brini||No||179||?|
|2010-10-21||2010-04-16||add7fe0||Updated the man page to reflect the behavioural change of create_temp_file()||-||David Sommerseth||No||185||3|
|2010-10-21||2010-03-10||1e02046||On TARGET_LINUX define _GNU_SOURCE if not defined||-||David Sommerseth||No||221||?|
|2011-04-24||2010-03-07||94d50a1||when deleting a route on win32, also add gateway address (otherwise netsh.exe will succeed, but silently ignore request)||-||Gert Doering||No||47||-|
|2010-10-21||2010-03-02||d04b858||Several updates to openvpn.8 (man page updates)||-||Karl O. Pinc||No||229||?|
|2011-03-25||2010-02-20||a698464||* undo mroute.c changes related to ipv6 payload, nothing to do w/ipv6 transport afterall.||-||JuanJo? Ciarlante||No||35||-|
|2010-10-21||2010-02-19||adfe37f||verb 5 logging wrongly reports received bytes||-||David Sommerseth||No||242||240|
|2010-10-21||2010-02-18||058f3d0||[PATCH] Change verify-cn so cn is no longer hardcoded in openvpn's config file||-||Karl O. Pinc||No||243||243|
|2011-03-25||2009-10-05||e293510||* TODO.ipv6 update||-||JuanJo? Ciarlante||No||130||-|
|2011-03-25||2009-09-24||e478770||* make possible to x-compile openvpn/win32 in Linux||-||JuanJo? Ciarlante||No||141||-|
For comparison, let's look at two big patchsets:
- PolarSSL support (~115 patches)
- total of 8 changes were asked for
- 3 of these seem trivial stylistic issues/typos
- Buildsystem revolution (52 patches)
- a couple of changes were asked for (e.g. to enable lzo by default)
Although this review was fairly limited in scope (~10% of 634 patches), some interesting things surfaced:
- A patch was modified during review only on two occasions (0,32% of all patches). That said, it was often difficult to know whether a patch was modified before being introduced. This could have had happened with patches that did not go through the usual mailing list scrutiny, see "?" in "Delay (after ACK)" column.
- The larger the patchset, the longer it takes to get merged (e.g. ~110 days for PolarSSL patches). Although this is a no-brainer, there's probably lots of room for improvement.
- Quite a few patches already go into Git without review, see "-" (i.e. "not applicable") in "Delay (after ACK)" column.
- The majority of ACKs were given by only 3 persons.
- It's often very difficult to find discussion about patches: see "?" in "Delay (after ACK)" column.