summaryrefslogtreecommitdiff
path: root/HACKING
blob: b9f96d637d6c6abd04e76f9e8fd28f66444ffc4b (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
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
     Rules for commits on the xmlsec module
     =========================================

0) DO NOT COMMIT DIRECTLY !
If you have a patch send a mail to xmlsec@aleksey.com mailing 
list (you must be subscribed to the list, go to 
http://www.aleksey.com/mailman/listinfo/xmlsec to subscribe).

If there is a problem in xmlsec module that prevents you
from building other major components then feel free to patch
first and then send a mail. This is an EXCEPTIONAL case and
you should be VERY carefull when you are doing this.

Igor Zlatkovic get an exception for the send before commit rule.

1) Coding style.
    - Formatting. Just for clarification, the formating is:

	tab size=8;indentation=4;insert spaces=yes 

    - Use explicit "!= NULL", "!= 0", etc. This makes code
    easier to read and remove warnings on some platform. 
    Example:
	BAD:
	    if(a) 
	GOOD:
	    if(a != NULL)
	or 
	    if(a != 0)
   
   - Put figure brackets '{}' even if you have only one operator
   in "if", "for", etc. This also makes code easier to read and 
   saves a lot of time when you need to quickly change something. 
   Example:
	BAD:
             if(a != NULL) 
		xmlFree(a);
	GOOD:
             if(a != NULL) {
		xmlFree(a);
	    }
    
    - Use round brackets '()' in conditions to show the precedence order.
    I don't remember what goes first '<<' or '*', do you?
    Example:
	BAD:
            if(privkey == NULL || pubkey == NULL)
	GOOD:
	    if((privkey == NULL) || (pubkey == NULL))
   
   - Use round brackets '()' for "return". 
   Example:
	BAD:
	    return 0;
	GOOD:
	    return(0);
    
    - Check for warnings! Use "--enable-pedantic" option
    for "configure.in" script to enable as much warnings as possible.
    Your patch should produce no new warnings and if you'll
    see something that you can fix, then do it.
    
    - Check for memory leaks. There is a built in support for 
    valgrind (http://devel-home.kde.org/~sewardj/). In order to use it,
    use "enable_static_linking" option for "configure.in" script to 
    force static linking of xmlsec command line utility and run 
    "make memcheck" from the top xmlsec source folder. The results are printed
    at the end. More detailed logs could be found in /tmp/test*.log files.

2) Coding practice
    - You should trust nobody! Anyone can fool you: user or another application
    might provide you incorrect data; call to xmlsec or system function might 
    fail with an error code; worse, the same call might fail but the return 
    code is "success" and so on. The patch fixes a lot of places where the 
    original code failed to check input data or function return values. 
    One of my favorite examples is the code that *silently* assumed that 
    base64 decoded value of a RSA public exponent obtained from XML fits 
    in a DWORD. And after that the code did memcpy to copy from xmlSecBuffer 
    to a DWORD variable *without* checking how much data are actualy copied! 
    The trivial DoS attack (at least DoS!!!) is to put very long base64 string
    in XML file and enjoy the server crash.
    One of the strongest sides of xmlsec library is that there are very few 
    known ways to crash it (and all of them are related to running the 
    application in an environment with a very limited memory to force a malloc 
    failure). To be a little paranoid is good in this context :)
    
    - malloc/free vs. xmlMalloc/xmlFree
    xmlsec library use libxml2 memory management functions. This provides an 
    easy way to replace default memory management functions with custom ones. 
    And this might be very usefull in some cases.
    Note that crypto library might use a different memory management
    functions! Be very carefully to do not mix them (i.e. get memory
    allocated by crypto library function and free it with xmFree).

    - Errors reporting (XMLSEC_ERRORS_R_XMLSEC_FAILED vs. XMLSEC_ERRORS_R_CRYPTO_FAILED)
    The correct usage rule is:
	if the failed function starts with "xmlSec" then use
    	    XMLSEC_ERRORS_R_XMLSEC_FAILED
	else if it is xmlMalloc/xmlFree/xmlStrdup/etc then use
            XMLSEC_ERRORS_R_MALLOC_FAILED
	else if the function starts with "xml" or "xslt" (i.e. it comes 
	from libxml or libxslt) then use
	    XMLSEC_ERRORS_R_XML_FAILED
	else if it is related to IO (fopen, fread, fwrite, etc.) then use
	    XMLSEC_ERRORS_R_IO_FAILED
	else if the function could be used only from xmlsec-crypto (i.e. 
	it is crypto engine related) then use
	    XMLSEC_ERRORS_R_CRYPTO_FAILED
	else if there is another reason (invalid data, invalid size, etc.)
	    corresponding error reason should be used
	else
	    it is something new and should be discussed
	fi						    
    Correct error reason is very important. For example,  some applications 
    ignore all the XMLSEC_ERRORS_R_XMLSEC_FAILED errors to get to the bottom of
    the errors stack and report the actual problem.
							    
    - Errors reporting: "size=%d;error=%d" instead of "size %d, error: %d":
    It would be great if xmlsec-crypto libraries can follow the error message 
    standard adopted in the other files of xmlsec library: 
	"<name1>=<value1>;<name2>=<value2>;..."
    This greatly helps when one needs to write a logs parser. For example, to 
    find the reason of memory allocation failures.				    

3) Preparing and submitting a patch.
If you want to submit a patch please create a pull request on GitHub and then
   send your pull request along with a short description of the problem or feature
    you are fixing/implementing to the xmlsec@aleksey.com mailing list 
    (you must be subscribed to the list, go to http://www.aleksey.com/mailman/listinfo/xmlsec to subscribe).
    If you are fixing a bug, it might be a good idea to create a GitHub ticket first
    (http://www.aleksey.com/xmlsec/bugs.html) for the record.
    
4) Building a release
- Cleanup, make sure no other changes are pending
  - make distclean
  - git status 
- Update Changelog
- Write about release changes in the release
  - docs/index.html and docs/news.html
- Update release number in 
  - configure.in (2 places at the top)
  - docs/download.html
- Create build
  - ./autogen.sh
  - make
- Build docs (watch for errors!)
  - make docs
- Commit the "prepare for X.Y.Z" release
  - git commit -m"prepare for X.Y.Z release" -a
- Run tests, make sure everything is OK
  - make check
- Build release
  - sudo ./scripts/build_release.sh
- Extract tar file, make sure it works
  - cd /tmp
  - tar xvfz /usr/src/redhat/SOURCE/xmlsec1-X.Y.z.tar.gz 
  - cd xmlsec1-X.Y.z
  - ./configure
  - make
  - make check
- Copy tar file to FTP/Web Download
- Copy docs/ folder to Web folder
- Write an announcement email to xmlsec@aleksey.com
- Update freshmeat.net
- Relax