forked from haproxy/haproxy
-
Notifications
You must be signed in to change notification settings - Fork 0
/
CONTRIBUTING
1020 lines (827 loc) · 54.5 KB
/
CONTRIBUTING
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
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
HOW TO GET YOUR CODE ACCEPTED IN HAPROXY
READ THIS CAREFULLY BEFORE SUBMITTING CODE
THIS DOCUMENT PROVIDES SOME RULES TO FOLLOW WHEN SENDING CONTRIBUTIONS. PATCHES
NOT FOLLOWING THESE RULES WILL SIMPLY BE IGNORED IN ORDER TO PROTECT ALL OTHER
RESPECTFUL CONTRIBUTORS' VALUABLE TIME.
Abstract
--------
If you have never contributed to HAProxy before, or if you did so and noticed
that nobody seems to be interested in reviewing your submission, please do read
this long document carefully. HAProxy maintainers are particularly demanding on
respecting certain simple rules related to general code and documentation style
as well as splitting your patches and providing high quality commit messages.
The reason behind this is that your patch will be met multiple times in the
future, when doing some backporting work or when bisecting a bug, and it is
critical that anyone can quickly decide if the patch is right, wrong, if it
misses something, if it must be reverted or needs to be backported. Maintainers
are generally benevolent with newcomers and will help them provided their work
indicates they have at least read this document. Some have improved over time,
to the point of being totally trusted and gaining commit access so they don't
need to depend on anyone to pick their code. On the opposite, those who insist
not making minimal efforts however will simply be ignored.
Background
----------
HAProxy is a community-driven project. But like most highly technical projects
it takes a lot of time to develop the skills necessary to be autonomous in the
project, and there is a very small core team helped by a small set of very
active participants. While most of the core team members work on the code as
part of their day job, most participants do it on a voluntary basis during
their spare time. The ideal model for developers is to spend their time:
1) developing new features
2) fixing bugs
3) doing maintenance backports
4) reviewing other people's code
It turns out that on a project like HAProxy, like many other similarly complex
projects, the time spent is exactly the opposite:
1) reviewing other people's code
2) doing maintenance backports
3) fixing bugs
4) developing new features
A large part of the time spent reviewing code often consists in giving basic
recommendations that are already explained in this file. In addition to taking
time, it is not appealing for people willing to spend one hour helping others
to do the same thing over and over instead of discussing the code design, and
it tends to delay the start of code reviews.
Regarding backports, they are necessary to provide a set of stable branches
that are deployed in production at many places. Load balancers are complex and
new features often induce undesired side effects in other areas, which we will
call bugs. Thus it's common for users to stick to a branch featuring everything
they need and not to upgrade too often. This backporting job is critical to the
ecosystem's health and must be done regularly. Very often the person devoting
some time on backports has little to no information about the relevance (let
alone importance) of a patch and is unlikely to be an expert in the area
affected by the patch. It's the role of the commit message to explain WHAT
problem the patch tries to solve, WHY it is estimated that it is a problem, and
HOW it tries to address it. With these elements, the person in charge of the
backports can decide whether or not to pick the patch. And if the patch does
not apply (which is common for older versions) they have information in the
commit message about the principle and choices that the initial developer made
and will try to adapt the patch sticking to these principles. Thus, the time
spent backporting patches solely depends on the code quality and the commit
message details and accuracy.
When it turns to fixing bugs, before declaring a bug, there is an analysis
phase. It starts with "is this behaviour expected", "is it normal", "under what
circumstances does it happen", "when did it start to happen", "was it intended",
"was it just overlooked", and "how to fix it without breaking the initial
intent". A utility called "git bisect" is usually involved in determining when
the behaviour started to happen. It determines the first patch which introduced
the new behaviour. If the patch is huge, touches many areas, is really difficult
to read because it needlessly reindents code or adds/removes line breaks out of
context, it will be very difficult to figure what part of this patch broke the
behaviour. Then once the part is figured, if the commit message doesn't provide
a detailed description about the intent of the patch, i.e. the problem it was
trying to solve, why and how, the developer landing on that patch will really
feel powerless. And very often in this case, the fix for the problem will break
something else or something that depended on the original patch.
But contrary to what it could look like, providing great quality patches is not
difficult, and developers will always help contributors improve their patches
quality because it's in their interest as well. History has shown that first
time contributors can provide an excellent work when they have carefully read
this document, and that people coming from projects with different practices
can grow from first-time contributor to trusted committer in about 6 months.
Preparation
-----------
It is possible that you'll want to add a specific feature to satisfy your needs
or one of your customers'. Contributions are welcome, however maintainers are
often very picky about changes. Patches that change massive parts of the code,
or that touch the core parts without any good reason will generally be rejected
if those changes have not been discussed first.
The proper place to discuss your changes is the HAProxy Mailing List. There are
enough skilled readers to catch hazardous mistakes and to suggest improvements.
There is no other place where you'll find as many skilled people on the project,
and these people can help you get your code integrated quickly. You can
subscribe to it by sending an empty e-mail at the following address :
haproxy+subscribe@formilux.org
It is not even necessary to subscribe, you can post there and verify via the
public list archives that your message was properly delivered. In this case you
should indicate in your message that you'd like responders to keep you CCed.
Please visit http://haproxy.org/ to figure available options to join the list.
If you have an idea about something to implement, *please* discuss it on the
list first. It has already happened several times that two persons did the same
thing simultaneously. This is a waste of time for both of them. It's also very
common to see some changes rejected because they're done in a way that will
conflict with future evolutions, or that does not leave a good feeling. It's
always unpleasant for the person who did the work, and it is unpleasant in
general because people's time and efforts are valuable and would be better
spent working on something else. That would not happen if these were discussed
first. There is no problem posting work in progress to the list, it happens
quite often in fact. Just prefix your mail subject with "RFC" (it stands for
"request for comments") and everyone will understand you'd like some opinion
on your work in progress. Also, don't waste your time with the doc when
submitting patches for review, only add the doc with the patch you consider
ready to merge (unless you need some help on the doc itself, of course).
Another important point concerns code portability. HAProxy requires gcc as the
C compiler, and may or may not work with other compilers. However it's known to
build using gcc 2.95 or any later version. As such, it is important to keep in
mind that certain facilities offered by recent versions must not be used in the
code:
- declarations mixed in the code (requires gcc >= 3.x and is a bad practice)
- GCC builtins without checking for their availability based on version and
architecture ;
- assembly code without any alternate portable form for other platforms
- use of stdbool.h, "bool", "false", "true" : simply use "int", "0", "1"
- in general, anything which requires C99 (such as declaring variables in
"for" statements)
Since most of these restrictions are just a matter of coding style, it is
normally not a problem to comply. Please read doc/coding-style.txt for all the
details.
When modifying some optional subsystem (SSL, Lua, compression, device detection
engines), please make sure the code continues to build (and to work) when these
features are disabled. Similarly, when modifying the SSL stack, please always
ensure that supported OpenSSL versions continue to build and to work, especially
if you modify support for alternate libraries. Clean support for the legacy
OpenSSL libraries is mandatory, support for its derivatives is a bonus and may
occasionally break even though a great care is taken. In other words, if you
provide a patch for OpenSSL you don't need to test its derivatives, but if you
provide a patch for a derivative you also need to test with OpenSSL.
If your work is very confidential and you can't publicly discuss it, you can
also mail willy@haproxy.org directly about it, but your mail may be waiting
several days in the queue before you get a response, if you get a response at
all. Retransmit if you don't get a response by one week. Please note that
direct sent e-mails to this address for non-confidential subjects may simply
be forwarded to the list or be deleted without notification. An auto-responder
bot is in place to try to detect e-mails from people asking for help and to
redirect them to the mailing list. Do not be surprised if this happens to you.
If you'd like a feature to be added but you think you don't have the skills to
implement it yourself, you should follow these steps :
1. discuss the feature on the mailing list. It is possible that someone
else has already implemented it, or that someone will tell you how to
proceed without it, or even why not to do it. It is also possible that
in fact it's quite easy to implement and people will guide you through
the process. That way you'll finally have YOUR patch merged, providing
the feature YOU need.
2. if you really can't code it yourself after discussing it, then you may
consider contacting someone to do the job for you. Some people on the
list might sometimes be OK with trying to do it.
The version control system used by the project (Git) keeps authorship
information in the form of the patch author's e-mail address. This way you will
be credited for your work in the project's history. If you contract with
someone to implement your idea you may have to discuss such modalities with
the person doing the work as by default this person will be mentioned as the
work's author.
Rules: the 12 laws of patch contribution
----------------------------------------
People contributing patches must apply the following rules. That may sound heavy
at the beginning but it's common sense more than anything else and contributors
do not think about them anymore after a few patches.
1) Comply with the license
Before modifying some code, you have read the LICENSE file ("main license")
coming with the sources, and all the files this file references. Certain
files may be covered by different licenses, in which case it will be
indicated in the files themselves. In any case, you agree to respect these
licenses and to contribute your changes under the same licenses. If you want
to create new files, they will be under the main license, or any license of
your choice that you have verified to be compatible with the main license,
and that will be explicitly mentioned in the affected files. The project's
maintainers are free to reject contributions proposing license changes they
feel are not appropriate or could cause future trouble.
2) Develop on development branch, not stable ones
Your work may only be based on the latest development version. No development
is made on a stable branch. If your work needs to be applied to a stable
branch, it will first be applied to the development branch and only then will
be backported to the stable branch. You are responsible for ensuring that
your work correctly applies to the development version. If at any moment you
are going to work on restructuring something important which may impact other
contributors, the rule that applies is that the first sent is the first
served. However it is considered good practice and politeness to warn others
in advance if you know you're going to make changes that may force them to
re-adapt their code, because they did probably not expect to have to spend
more time discovering your changes and rebasing their work.
3) Read and respect the coding style
You have read and understood "doc/coding-style.txt", and you're actively
determined to respect it and to enforce it on your coworkers if you're going
to submit a team's work. We don't care what text editor you use, whether it's
an hex editor, cat, vi, emacs, Notepad, Word, or even Eclipse. The editor is
only the interface between you and the text file. What matters is what is in
the text file in the end. The editor is not an excuse for submitting poorly
indented code, which only proves that the person has no consideration for
quality and/or has done it in a hurry (probably worse). Please note that most
bugs were found in low-quality code. Reviewers know this and tend to be much
more reluctant to accept poorly formatted code because by experience they
won't trust their author's ability to write correct code. It is also worth
noting that poor quality code is painful to read and may result in nobody
willing to waste their time even reviewing your work.
4) Present clean work
The time it takes for you to polish your code is always much smaller than the
time it takes others to do it for you, because they always have to wonder if
what they see is intended (meaning they didn't understand something) or if it
is a mistake that needs to be fixed. And since there are less reviewers than
submitters, it is vital to spread the effort closer to where the code is
written and not closer to where it gets merged. For example if you have to
write a report for a customer that your boss wants to review before you send
it to the customer, will you throw on his desk a pile of paper with stains,
typos and copy-pastes everywhere ? Will you say "come on, OK I made a mistake
in the company's name but they will find it by themselves, it's obvious it
comes from us" ? No. When in doubt, simply ask for help on the mailing list.
5) Documentation is very important
There are four levels of importance of quality in the project :
- The most important one, and by far, is the quality of the user-facing
documentation. This is the first contact for most users and it immediately
gives them an accurate idea of how the project is maintained. Dirty docs
necessarily belong to a dirty project. Be careful to the way the text you
add is presented and indented. Be very careful about typos, usual mistakes
such as double consonants when only one is needed or "it's" instead of
"its", don't mix US English and UK English in the same paragraph, etc.
When in doubt, check in a dictionary. Fixes for existing typos in the doc
are always welcome and chasing them is a good way to become familiar with
the project and to get other participants' respect and consideration.
- The second most important level is user-facing messages emitted by the
code. You must try to see all the messages your code produces to ensure
they are understandable outside of the context where you wrote them,
because the user often doesn't expect them. That's true for warnings, and
that's even more important for errors which prevent the program from
working and which require an immediate and well understood fix in the
configuration. It's much better to say "line 35: compression level must be
an integer between 1 and 9" than "invalid argument at line 35". In HAProxy,
error handling roughly represents half of the code, and that's about 3/4 of
the configuration parser. Take the time to do something you're proud of. A
good rule of thumb is to keep in mind that your code talks to a human and
tries to teach them how to proceed. It must then speak like a human.
- The third most important level is the code and its accompanying comments,
including the commit message which is a complement to your code and
comments. It's important for all other contributors that the code is
readable, fluid, understandable and that the commit message describes what
was done, the choices made, the possible alternatives you thought about,
the reason for picking this one and its limits if any. Comments should be
written where it's easy to have a doubt or after some error cases have been
wiped out and you want to explain what possibilities remain. All functions
must have a comment indicating what they take on input and what they
provide on output. Please adjust the comments when you copy-paste a
function or change its prototype, this type of lazy mistake is too common
and very confusing when reading code later to debug an issue. Do not forget
that others will feel really angry at you when they have to dig into your
code for a bug that your code caused and they feel like this code is dirty
or confusing, that the commit message doesn't explain anything useful and
that the patch should never have been accepted in the first place. That
will strongly impact your reputation and will definitely affect your
chances to contribute again!
- The fourth level of importance is in the technical documentation that you
may want to add with your code. Technical documentation is always welcome
as it helps others make the best use of your work and to go exactly in the
direction you thought about during the design. This is also what reduces
the risk that your design gets changed in the near future due to a misuse
and/or a poor understanding. All such documentation is actually considered
as a bonus. It is more important that this documentation exists than that
it looks clean. Sometimes just copy-pasting your draft notes in a file to
keep a record of design ideas is better than losing them. Please do your
best so that other ones can read your doc. If these docs require a special
tool such as a graphics utility, ensure that the file name makes it
unambiguous how to process it. So there are no rules here for the contents,
except one. Please write the date in your file. Design docs tend to stay
forever and to remain long after they become obsolete. At this point that
can cause harm more than it can help. Writing the date in the document
helps developers guess the degree of validity and/or compare them with the
date of certain commits touching the same area.
6) US-ASCII only!
All text files and commit messages are written using the US-ASCII charset.
Please be careful that your contributions do not contain any character not
printable using this charset, as they will render differently in different
editors and/or terminals. Avoid latin1 and more importantly UTF-8 which some
editors tend to abuse to replace some US-ASCII characters with their
typographic equivalent which aren't readable anymore in other editors. The
only place where alternative charsets are tolerated is in your name in the
commit message, but it's at your own risk as it can be mangled during the
merge. Anyway if you have an e-mail address, you probably have a valid
US-ASCII representation for it as well.
7) Comments
Be careful about comments when you move code around. It's not acceptable that
a block of code is moved to another place leaving irrelevant comments at the
old place, just like it's not acceptable that a function is duplicated without
the comments being adjusted. The example below started to become quite common
during the 1.6 cycle, it is not acceptable and wastes everyone's time :
/* Parse switching <str> to build rule <rule>. Returns 0 on error. */
int parse_switching_rule(const char *str, struct rule *rule)
{
...
}
/* Parse switching <str> to build rule <rule>. Returns 0 on error. */
void execute_switching_rule(struct rule *rule)
{
...
}
This patch is not acceptable either (and it's unfortunately not that rare) :
+ if (!session || !arg || list_is_empty(&session->rules->head))
+ return 0;
+
/* Check if session->rules is valid before dereferencing it */
if (!session->rules_allocated)
return 0;
- if (!arg || list_is_empty(&session->rules->head))
- return 0;
-
8) Short, readable identifiers
Limit the length of your identifiers in the code. When your identifiers start
to sound like sentences, it's very hard for the reader to keep on track with
what operation they are observing. Also long names force expressions to fit
on several lines which also cause some difficulties to the reader. See the
example below :
int file_name_len_including_global_path;
int file_name_len_without_global_path;
int global_path_len_or_zero_if_default;
if (global_path)
global_path_len_or_zero_if_default = strlen(global_path);
else
global_path_len_or_zero_if_default = 0;
file_name_len_without_global_path = strlen(file_name);
file_name_len_including_global_path =
file_name_len_without_global_path + 1 + /* for '/' */
global_path_len_or_zero_if_default ?
global_path_len_or_zero_if_default : default_path_len;
Compare it to this one :
int f, p;
p = global_path ? strlen(global_path) : default_path_len;
f = p + 1 + strlen(file_name); /* 1 for '/' */
A good rule of thumb is that if your identifiers start to contain more than
3 words or more than 15 characters, they can become confusing. For function
names it's less important especially if these functions are rarely used or
are used in a complex context where it is important to differentiate between
their multiple variants.
9) Unified diff only
The best way to build your patches is to use "git format-patch". This means
that you have committed your patch to a local branch, with an appropriate
subject line and a useful commit message explaining what the patch attempts
to do. It is not strictly required to use git, but what is strictly required
is to have all these elements in the same mail, easily distinguishable, and
a patch in "diff -up" format (which is also the format used by Git). This
means the "unified" diff format must be used exclusively, and with the
function name printed in the diff header of each block. That significantly
helps during reviews. Keep in mind that most reviews are done on the patch
and not on the code after applying the patch. Your diff must keep some
context (3 lines above and 3 lines below) so that there's no doubt where the
code has to be applied. Don't change code outside of the context of your
patch (eg: take care of not adding/removing empty lines once you remove
your debugging code). If you are using Git (which is strongly recommended),
always use "git show" after doing a commit to ensure it looks good, and
enable syntax coloring that will automatically report in red the trailing
spaces or tabs that your patch added to the code and that must absolutely be
removed. These ones cause a real pain to apply patches later because they
mangle the context in an invisible way. Such patches with trailing spaces at
end of lines will be rejected.
10) One patch per feature
Please cut your work in series of patches that can be independently reviewed
and merged. Each patch must do something on its own that you can explain to
someone without being ashamed of what you did. For example, you must not say
"This is the patch that implements SSL, it was tricky". There's clearly
something wrong there, your patch will be huge, will definitely break things
and nobody will be able to figure what exactly introduced the bug. However
it's much better to say "I needed to add some fields in the session to store
the SSL context so this patch does this and doesn't touch anything else, so
it's safe". Also when dealing with series, you will sometimes fix a bug that
one of your patches introduced. Please do merge these fixes (eg: using git
rebase -i and squash or fixup), as it is not acceptable to see patches which
introduce known bugs even if they're fixed later. Another benefit of cleanly
splitting patches is that if some of your patches need to be reworked after
a review, the other ones can still be merged so that you don't need to care
about them anymore. When sending multiple patches for review, prefer to send
one e-mail per patch than all patches in a single e-mail. The reason is that
not everyone is skilled in all areas nor has the time to review everything
at once. With one patch per e-mail, it's easy to comment on a single patch
without giving an opinion on the other ones, especially if a long thread
starts about one specific patch on the mailing list. "git send-email" does
that for you though it requires a few trials before getting it right.
If you can, please always put all the bug fixes at the beginning of the
series. This often makes it easier to backport them because they will not
depend on context that your other patches changed. As a hint, if you can't
do this, there are little chances that your bug fix can be backported.
11) Real commit messages please!
The commit message is how you're trying to convince a maintainer to adopt
your work and maintain it as long as possible. A dirty commit message almost
always comes with dirty code. Too short a commit message indicates that too
short an analysis was done and that side effects are extremely likely to be
encountered. It's the maintainer's job to decide to accept this work in its
current form or not, with the known constraints. Some patches which rework
architectural parts or fix sensitive bugs come with 20-30 lines of design
explanations, limitations, hypothesis or even doubts, and despite this it
happens when reading them 6 months later while trying to identify a bug that
developers still miss some information about corner cases.
So please properly format your commit messages. To get an idea, just run
"git log" on the file you've just modified. Patches always have the format
of an e-mail made of a subject, a description and the actual patch. If you
are sending a patch as an e-mail formatted this way, it can quickly be
applied with limited effort so that's acceptable :
- A subject line (may wrap to the next line, but please read below)
- an empty line (subject delimiter)
- a non-empty description (the body of the e-mail)
- the patch itself
The subject describes the "What" of the change ; the description explains
the "why", the "how" and sometimes "what next". For example a commit message
looking like this will be rejected :
| From: Mr Foobar <foobar@example.com>
| Subject: BUG: fix typo in ssl_sock
|
This one as well (too long subject, not the right place for the details) :
| From: Mr Foobar <foobar@example.com>
| Subject: BUG/MEDIUM: ssl: use an error flag to prevent ssl_read() from
| returning 0 when dealing with large buffers because that can cause
| an infinite loop
|
This one ought to be used instead :
| From: Mr Foobar <foobar@example.com>
| Subject: BUG/MEDIUM: ssl: fix risk of infinite loop in ssl_sock
|
| ssl_read() must not return 0 on error or the caller may loop forever.
| Instead we add a flag to the connection to notify about the error and
| check it at all call places. This situation can only happen with large
| buffers so a workaround is to limit buffer sizes. Another option would
| have been to return -1 but it required to use signed ints everywhere
| and would have made the patch larger and riskier. This fix should be
| backported to versions 1.2 and upper.
It is important to understand that for any reader to guess the text above
when it's absent, it will take a huge amount of time. If you made the
analysis leading to your patch, you must explain it, including the ideas
you dropped if you had a good reason for this.
While it's not strictly required to use Git, it is strongly recommended
because it helps you do the cleanest job with the least effort. But if you
are comfortable with writing clean e-mails and inserting your patches, you
don't need to use Git.
But in any case, it is important that there is a clean description of what
the patch does, the motivation for what it does, why it's the best way to do
it, its impacts, and what it does not yet cover. And this is particularly
important for bugs. A patch tagged "BUG" must absolutely explain what the
problem is, why it is considered as a bug. Anybody, even non-developers,
should be able to tell whether or not a patch is likely to address an issue
they are facing. Indicating what the code will do after the fix doesn't help
if it does not say what problem is encountered without the patch. Note that
in some cases the bug is purely theoretical and observed by reading the code.
In this case it's perfectly fine to provide an estimate about possible
effects. Also, in HAProxy, like many projects which take a great care of
maintaining stable branches, patches are reviewed later so that some of them
can be backported to stable releases.
While reviewing hundreds of patches can seem cumbersome, with a proper
formatting of the subject line it actually becomes very easy. For example,
here's how one can find patches that need to be reviewed for backports (bugs
and doc) between since commit ID 827752e :
$ git log --oneline 827752e.. | grep 'BUG\|DOC'
0d79cf6 DOC: fix function name
bc96534 DOC: ssl: missing LF
10ec214 BUG/MEDIUM: lua: the lua function Channel:close() causes a segf
bdc97a8 BUG/MEDIUM: lua: outgoing connection was broken since 1.6-dev2
ba56d9c DOC: mention support for RFC 5077 TLS Ticket extension in start
f1650a8 DOC: clarify some points about SSL and the proxy protocol
b157d73 BUG/MAJOR: peers: fix current table pointer not re-initialized
e1ab808 BUG/MEDIUM: peers: fix wrong message id on stick table updates
cc79b00 BUG/MINOR: ssl: TLS Ticket Key rotation broken via socket comma
d8e42b6 DOC: add new file intro.txt
c7d7607 BUG/MEDIUM: lua: bad error processing
386a127 DOC: match several lua configuration option names to those impl
0f4eadd BUG/MEDIUM: counters: ensure that src_{inc,clr}_gpc0 creates a
It is made possible by the fact that subject lines are properly formatted and
always respect the same principle : one part indicating the nature and
severity of the patch, another one to indicate which subsystem is affected,
and the last one is a succinct description of the change, with the important
part at the beginning so that it's obvious what it does even when lines are
truncated like above. The whole stable maintenance process relies on this.
For this reason, it is mandatory to respect some easy rules regarding the
way the subject is built. Please see the section below for more information
regarding this formatting.
As a rule of thumb, your patch MUST NEVER be made only of a subject line,
it *must* contain a description. Even one or two lines, or indicating
whether a backport is desired or not. It turns out that single-line commits
are so rare in the Git world that they require special manual (hence
painful) handling when they are backported, and at least for this reason
it's important to keep this in mind.
Maintainers who pick your patch may slightly adjust the description as they
see fit. Do not see this as a failure to do a clean job, it just means they
think it will help them do their daily job this way. The code may also be
slightly adjusted before being merged (non-functional changes only, fix for
typos, tabs vs spaces for example), unless your patch contains a
Signed-off-By tag, in which case they will either modify it and mention the
changes after your Signed-off-By line, or (more likely) ask you to perform
these changes yourself. This ability to slightly adjust a patch before
merging is is the main reason for not using pull requests which do not
provide this facility and will require to iterate back and forth with the
submitter and significantly delay the patch inclusion.
Each patch fixing a bug MUST be tagged with "BUG", a severity level, an
indication of the affected subsystem and a brief description of the nature
of the issue in the subject line, and a detailed analysis in the message
body. The explanation of the user-visible impact and the need for
backporting to stable branches or not are MANDATORY. Bug fixes with no
indication will simply be rejected as they are very likely to cause more
harm when nobody is able to tell whether or not the patch needs to be
backported or can be reverted in case of regression.
When fixing a bug which is reproducible, if possible, the contributors are
strongly encouraged to write a regression testing VTC file for varnishtest
to add to reg-tests directory. More information about varnishtest may be
found in README file of reg-tests directory and in doc/regression-testing.txt
file.
12) Discuss on the mailing list
Note, some first-time contributors might feel impressed or scared by posting
to a list. This list is frequented only by nice people who are willing to
help you polish your work so that it is perfect and can last long. What you
think could be perceived as a proof of incompetence or lack of care will
instead be a proof of your ability to work with a community. You will not be
judged nor blamed for making mistakes. The project maintainers are the ones
creating the most bugs and mistakes anyway, and nobody knows the project in
its entirety anymore so you're just like anyone else. And people who have no
consideration for other's work are quickly ejected from the list so the
place is as safe and welcoming to new contributors as it is to long time
ones.
When submitting changes, please always CC the mailing list address so that
everyone gets a chance to spot any issue in your code. It will also serve
as an advertisement for your work, you'll get more testers quicker and
you'll feel better knowing that people really use your work. It's often
convenient to prepend "[PATCH]" in front of your mail's subject to mention
that this e-mail contains a patch (or a series of patches), because it will
easily catch reviewer's attention. It's automatically done by tools such as
"git format-patch" and "git send-email". If you don't want your patch to be
merged yet and prefer to show it for discussion, better tag it as "[RFC]"
(stands for "Request For Comments") and it will be reviewed but not merged
without your approval. It is also important to CC any author mentioned in
the file you change, or a subsystem maintainers whose address is mentioned
in a MAINTAINERS file. Not everyone reads the list on a daily basis so it's
very easy to miss some changes. Don't consider it as a failure when a
reviewer tells you you have to modify your patch, actually it's a success
because now you know what is missing for your work to get accepted. That's
why you should not hesitate to CC enough people. Don't copy people who have
no deal with your work area just because you found their address on the
list. That's the best way to appear careless about their time and make them
reject your changes in the future.
Patch classifying rules
-----------------------
There are 3 criteria of particular importance in any patch :
- its nature (is it a fix for a bug, a new feature, an optimization, ...)
- its importance, which generally reflects the risk of merging/not merging it
- what area it applies to (eg: http, stats, startup, config, doc, ...)
It's important to make these 3 criteria easy to spot in the patch's subject,
because it's the first (and sometimes the only) thing which is read when
reviewing patches to find which ones need to be backported to older versions.
It also helps when trying to find which patch is the most likely to have caused
a regression.
Specifically, bugs must be clearly easy to spot so that they're never missed.
Any patch fixing a bug must have the "BUG" tag in its subject. Most common
patch types include :
- BUG fix for a bug. The severity of the bug should also be indicated
when known. Similarly, if a backport is needed to older versions,
it should be indicated on the last line of the commit message. The
commit message MUST ABSOLUTELY describe the problem and its impact
to non-developers. Any user must be able to guess if this patch is
likely to fix a problem they are facing. Even if the bug was
discovered by accident while reading the code or running an
automated tool, it is mandatory to try to estimate what potential
issue it might cause and under what circumstances. There may even
be security implications sometimes so a minimum analysis is really
required. Also please think about stable maintainers who have to
build the release notes, they need to have enough input about the
bug's impact to explain it. If the bug has been identified as a
regression brought by a specific patch or version, this indication
will be appreciated too. New maintenance releases are generally
emitted when a few of these patches are merged. If the bug is a
vulnerability for which a CVE identifier was assigned before you
publish the fix, you can mention it in the commit message, it will
help distro maintainers.
- CLEANUP code cleanup, silence of warnings, etc... theoretically no impact.
These patches will rarely be seen in stable branches, though they
may appear when they remove some annoyance or when they make
backporting easier. By nature, a cleanup is always of minor
importance and it's not needed to mention it.
- DOC updates to any of the documentation files, including README. Many
documentation updates are backported since they don't impact the
product's stability and may help users avoid bugs. So please
indicate in the commit message if a backport is desired. When a
feature gets documented, it's preferred that the doc patch appears
in the same patch or after the feature patch, but not before, as it
becomes confusing when someone working on a code base including
only the doc patch won't understand why a documented feature does
not work as documented.
- REORG code reorganization. Some blocks may be moved to other places,
some important checks might be swapped, etc... These changes
always present a risk of regression. For this reason, they should
never be mixed with any bug fix nor functional change. Code is
only moved as-is. Indicating the risk of breakage is highly
recommended. Minor breakage is tolerated in such patches if trying
to fix it at once makes the whole change even more confusing. That
may happen for example when some #ifdefs need to be propagated in
every file consecutive to the change.
- BUILD updates or fixes for build issues. Changes to makefiles also fall
into this category. The risk of breakage should be indicated if
known. It is also appreciated to indicate what platforms and/or
configurations were tested after the change.
- OPTIM some code was optimised. Sometimes if the regression risk is very
low and the gains significant, such patches may be merged in the
stable branch. Depending on the amount of code changed or replaced
and the level of trust the author has in the change, the risk of
regression should be indicated. If the optimization depends on the
architecture or on build options, it is important to verify that
the code continues to work without it.
- RELEASE release of a new version (development or stable).
- LICENSE licensing updates (may impact distro packagers).
- REGTEST updates to any of the regression testing files found in reg-tests
directory, including README or any documentation file.
When the patch cannot be categorized, it's best not to put any type tag, and to
only use a risk or complexity information only as below. This is commonly the
case for new features, which development versions are mostly made of.
The importance, complexity of the patch, or severity of the bug it fixes must
be indicated when relevant. A single upper-case word is preferred, among :
- MINOR minor change, very low risk of impact. It is often the case for
code additions that don't touch live code. As a rule of thumb, a
patch tagged "MINOR" is safe enough to be backported to stable
branches. For a bug, it generally indicates an annoyance, nothing
more.
- MEDIUM medium risk, may cause unexpected regressions of low importance or
which may quickly be discovered. In short, the patch is safe but
touches working areas and it is always possible that you missed
something you didn't know existed (eg: adding a "case" entry or
an error message after adding an error code to an enum). For a bug,
it generally indicates something odd which requires changing the
configuration in an undesired way to work around the issue.
- MAJOR major risk of hidden regression. This happens when large parts of
the code are rearranged, when new timeouts are introduced, when
sensitive parts of the session scheduling are touched, etc... We
should only exceptionally find such patches in stable branches when
there is no other option to fix a design issue. For a bug, it
indicates severe reliability issues for which workarounds are
identified with or without performance impacts.
- CRITICAL medium-term reliability or security is at risk and workarounds,
if they exist, might not always be acceptable. An upgrade is
absolutely required. A maintenance release may be emitted even if
only one of these bugs are fixed. Note that this tag is only used
with bugs. Such patches must indicate what is the first version
affected, and if known, the commit ID which introduced the issue.
The expected length of the commit message grows with the importance of the
change. While a MINOR patch may sometimes be described in 1 or 2 lines, MAJOR
or CRITICAL patches cannot have less than 10-15 lines to describe exactly the
impacts otherwise the submitter's work will be considered as rough sabotage.
If you are sending a new patch series after a review, it is generally good to
enumerate at the end of the commit description what changed from the previous
one as it helps reviewers quickly glance over such changes and not re-read the
rest.
For BUILD, DOC and CLEANUP types, this tag is not always relevant and may be
omitted.
The area the patch applies to is quite important, because some areas are known
to be similar in older versions, suggesting a backport might be desirable, and
conversely, some areas are known to be specific to one version. The area is a
single-word lowercase name the contributor find clear enough to describe what
part is being touched. The following list of tags is suggested but not
exhaustive:
- examples example files. Be careful, sometimes these files are packaged.
- tests regression test files. No code is affected, no need to upgrade.
- reg-tests regression test files for varnishtest. No code is affected, no
need to upgrade.
- init initialization code, arguments parsing, etc...
- config configuration parser, mostly used when adding new config keywords
- http the HTTP engine
- stats the stats reporting engine
- cli the stats socket CLI
- checks the health checks engine (eg: when adding new checks)
- sample the sample fetch system (new fetch or converter functions)
- acl the ACL processing core or some ACLs from other areas
- filters everything related to the filters core
- peers the peer synchronization engine
- lua the Lua scripting engine
- listeners everything related to incoming connection settings
- frontend everything related to incoming connection processing
- backend everything related to LB algorithms and server farm
- session session processing and flags (very sensible, be careful)
- server server connection management, queueing
- spoe SPOE code
- ssl the SSL/TLS interface
- proxy proxy maintenance (start/stop)
- log log management
- poll any of the pollers
- halog the halog sub-component in the admin directory
- htx general HTX subsystem
- mux-h1 HTTP/1.x multiplexer/demultiplexer
- mux-h2 HTTP/2 multiplexer/demultiplexer
- h1 general HTTP/1.x protocol parser
- h2 general HTTP/2 protocol parser
Other names may be invented when more precise indications are meaningful, for
instance : "cookie" which indicates cookie processing in the HTTP core. Last,
indicating the name of the affected file is also a good way to quickly spot
changes. Many commits were already tagged with "stream_sock" or "cfgparse" for
instance.
It is required that the type of change and the severity when relevant are
indicated, as well as the touched area when relevant as well in the patch
subject. Normally, we would have the 3 most often. The two first criteria should
be present before a first colon (':'). If both are present, then they should be
delimited with a slash ('/'). The 3rd criterion (area) should appear next, also
followed by a colon. Thus, all of the following subject lines are valid :
Examples of subject lines :
- DOC: document options forwardfor to logasap
- DOC/MAJOR: reorganize the whole document and change indenting
- BUG: stats: connection reset counters must be plain ascii, not HTML
- BUG/MINOR: stats: connection reset counters must be plain ascii, not HTML
- MEDIUM: checks: support multi-packet health check responses
- RELEASE: Released version 1.4.2
- BUILD: stats: stdint is not present on solaris
- OPTIM/MINOR: halog: make fgets parse more bytes by blocks
- REORG/MEDIUM: move syscall redefinition to specific places
Please do not use square brackets anymore around the tags, because they induce
more work when merging patches, which need to be hand-edited not to lose the
enclosed part.
In fact, one of the only square bracket tags that still makes sense is '[RFC]'
at the beginning of the subject, when you're asking for someone to review your
change before getting it merged. If the patch is OK to be merged, then it can
be merge as-is and the '[RFC]' tag will automatically be removed. If you don't
want it to be merged at all, you can simply state it in the message, or use an
alternate 'WIP/' prefix in front of your tag tag ("work in progress").
The tags are not rigid, follow your intuition first, and they may be readjusted
when your patch is merged. It may happen that a same patch has a different tag
in two distinct branches. The reason is that a bug in one branch may just be a
cleanup or safety measure in the other one because the code cannot be triggered.
Working with Git
----------------
For a more efficient interaction between the mainline code and your code, you
are strongly encouraged to try the Git version control system :
http://git-scm.com/
It's very fast, lightweight and lets you undo/redo your work as often as you
want, without making your mistakes visible to the rest of the world. It will
definitely help you contribute quality code and take other people's feedback
in consideration. In order to clone the HAProxy Git repository :
$ git clone http://git.haproxy.org/git/haproxy.git/ (development)
If you decide to use Git for your developments, then your commit messages will
have the subject line in the format described above, then the whole description
of your work (mainly why you did it) will be in the body. You can directly send
your commits to the mailing list, the format is convenient to read and process.
It is recommended to create a branch for your work that is based on the master
branch :
$ git checkout -b 20150920-fix-stats master
You can then do your work and even experiment with multiple alternatives if you
are not completely sure that your solution is the best one :
$ git checkout -b 20150920-fix-stats-v2
Then reorder/merge/edit your patches :
$ git rebase -i master
When you think you're ready, reread your whole patchset to ensure there is no
formatting or style issue :
$ git show master..
And once you're satisfied, you should update your master branch to be sure that
nothing changed during your work (only needed if you left it unattended for days
or weeks) :
$ git checkout -b 20150920-fix-stats-rebased
$ git fetch origin master:master
$ git rebase master
You can build a list of patches ready for submission like this :
$ git format-patch master
The output files are the patches ready to be sent over e-mail, either via a
regular e-mail or via git send-email (carefully check the man page). Don't
destroy your other work branches until your patches get merged, it may happen
that earlier designs will be preferred for various reasons. Patches should be
sent to the mailing list : haproxy@formilux.org and CCed to relevant subsystem
maintainers or authors of the modified files if their address appears at the
top of the file.
Please don't send pull requests, they are really inconvenient as they make it
much more complicate to perform minor adjustments, and nobody benefits from
any comment on the code while on a list all subscribers learn a little bit on
each review of anyone else's code.
What to do if your patch is ignored
-----------------------------------
All patches merged are acknowledged by the maintainer who picked it. If you
didn't get an acknowledgement, check the mailing list archives to see if your
mail was properly delivered there and possibly if anyone responded and you did
not get their response (please look at http://haproxy.org/ for the mailing list
archive's address).
If you see that your mail is there but nobody responded, please recheck:
- was the subject clearly indicating that it was a patch and/or that you were
seeking some review?
- was your email mangled by your mail agent? If so it's possible that
nobody had the willingness yet to mention it.
- was your email sent as HTML? If so it definitely ended in spam boxes
regardless of the archives.
- did the patch violate some of the principles explained in this document?
If none of these cases matches, it might simply be that everyone was busy when
your patch was sent and that it was overlooked. In this case it's fine to
either resubmit it or respond to your own email asking if anything's wrong
about it. In general don't expect a response after one week of silence, just
because your email will not appear in anyone else's current window. So after
one week it's time to resubmit.
Among the mistakes that tend to make reviewers not respond are those who send
multiple versions of a patch in a row. It's natural for others then to wait for
the series to stabilize. And once it doesn't move anymore everyone forgot about
it. As a rule of thumb, if you have to update your original email more than
twice, first double-check that your series is really ready for submission, and
second, start a new thread and stop responding to the previous one. In this
case it is well appreciated to mention a version of your patch set in the
subject such as "[PATCH v2]", so that reviewers can immediately spot the new
version and not waste their time on the old one.
If you still do not receive any response, it is possible that you've already
played your last card by not respecting the basic principles multiple times
despite being told about it several times, and that nobody is willing to spend
more of their time than normally needed with your work anymore. Your best
option at this point probably is to ask "did I do something wrong" than to
resend the same patches.
How to be sure to irritate everyone
-----------------------------------
Among the best ways to quickly lose everyone's respect, there is this small
selection, which should help you improve the way you work with others, if
you notice you're already practising some of them:
- repeatedly send improperly formatted commit messages, with no type or
severity, or with no commit message body. These ones require manual
edition, maintainers will quickly learn to recognize your name.
- repeatedly send patches which break something, and disappear or take a long
time to provide a fix.
- fail to respond to questions related to features you have contributed in
the past, which can further lead to the feature being declared unmaintained
and removed in a future version.