1------------------------------------------------------------------------------- 2HACKING 3------------------------------------------------------------------------------- 4 5Coding style 6------------ 7 8This project is programmed using the Linux kernel coding style: 9 10 https://www.kernel.org/doc/html/latest/process/coding-style.html 11 12Please use the same style for any code contributions, thanks! 13 14The Python decoders should follow the usual Python conventions and use 15Python idioms as far as it makes sense. The coding style should mostly follow 16the Python PEP-8, which includes the convention of 4 spaces for indentation: 17 18 http://www.python.org/dev/peps/pep-0008/ 19 20 21Contributions 22------------- 23 24 - In order to contribute you should ideally clone the git repository and 25 let us know (preferably via IRC, or via the mailing list) from where to 26 pull/review your changes. You can use github.com, or any other public git 27 hosting site. 28 29 - Alternatively, patches can be sent to the development mailinglist at 30 sigrok-devel@lists.sourceforge.net (please subscribe to the list first). 31 32 https://lists.sourceforge.net/lists/listinfo/sigrok-devel 33 34 35Random notes 36------------ 37 38 - Don't do variable declarations in compound statements, only at the 39 beginning of a function. 40 41 - Generally avoid assigning values to variables at declaration time, 42 especially so for complex and/or run-time dependent values. 43 44 - Consistently use g_*malloc() / g_*malloc0(). Do not use standard 45 malloc()/calloc() if it can be avoided (sometimes other libs such 46 as libftdi can return malloc()'d memory, for example). 47 48 - Always properly match allocations with the proper *free() functions. If 49 glib's g_*malloc()/g_*malloc0() was used, use g_free() to free the 50 memory. Otherwise use standard free(). Never use the wrong function! 51 52 - We assume that "small" memory allocations (< 1MB) will always succeed. 53 Thus, it's fine to use g_malloc() or g_malloc0() for allocations of 54 simple/small structs and such (instead of using g_try_malloc()), and 55 there's no need to check the return value. 56 57 Do use g_try_malloc() or g_try_malloc0() for large (>= 1MB) allocations 58 and check the return value. 59 60 - You should never print any messages (neither to stdout nor stderr nor 61 elsewhere) "manually" via e.g. printf() or g_log() or similar functions. 62 Only srd_err()/srd_warn()/srd_info()/srd_dbg()/srd_spew() should be used. 63 64 - Use glib's gboolean / TRUE / FALSE for boolean types consistently. 65 Do not use <stdbool.h> and its true / false, and do not invent private 66 definitions for this either. 67 68 - Consistently use the same naming convention for #include guards in headers: 69 <PROJECTNAME>_<PATH_TO_FILE>_<FILE> 70 This ensures that all #include guards are always unique and consistent. 71 Example: LIBSIGROKDECODE_LIBSIGROKDECODE_INTERNAL_H 72 73 - Consistently use the same naming convention for API functions: 74 <libprefix>_<groupname>_<action>(). 75 76 Examples: 77 srd_log_loglevel_set(), srd_log_loglevel_get(), srd_log_handler_set(), 78 srd_log_handler_set_default(), and so on. 79 80 Getter/setter function names should usually end with "_get" or "_set". 81 Functions creating new "objects" should end with "_new". 82 Functions destroying "objects" should end with "_destroy". 83 Functions adding or removing items (e.g. from lists) should end with 84 either "_add" or "_remove". 85 Functions operating on all items from a list (not on only one of them), 86 should end with "_all", e.g. "_remove_all", "_get_all", and so on. 87 Use "_remove_all" in favor of "_clear" for consistency. 88 89 - All enums should generally use an explicit start number of 10000. 90 If there are multiple "categories" in the enum entries, each category 91 should be 10000 entries apart from the next one. The start of categories 92 are thus 10000, 20000, 30000, and so on. 93 94 Adding items to an enum MUST always append to a "category", never add 95 items in the middle of a category. The order of items MUST NOT be changed. 96 Any of the above would break the ABI. 97 98 The enum item 0 is special and is used as terminator in some lists, thus 99 enums should not use this for "valid" entries (and start at 10000 instead). 100 101 102Doxygen 103------- 104 105 - In Doxygen comments, put an empty line between the block of @param lines 106 and the final @return line. The @param lines themselves (if there is more 107 than one) are not separated by empty lines. 108 109 - Mark private functions (SRD_PRIV) with /** @private */, so that Doxygen 110 doesn't include them in the output. Functions that are "static" anyway 111 don't need to be marked like this. 112 113 - Mark private variables/#defines with /** @cond PRIVATE */ and 114 /** @endcond */, so that Doxygen doesn't include them in the output. 115 Variables that are "static" don't need to be marked like this. 116 117 - Mark all public API functions (SRD_API) with a @since tag which indicates 118 in which release the respective function was added (e.g. "@since 0.1.0"). 119 120 If the function has existed before, but its API changed later, the @since 121 tag should mention only the release when the API last changed. 122 123 Example: The srd_foo() call was added in 0.1.0, but the API changed in 124 the later 0.2.0 release. The docs should read "@since 0.2.0" in that case. 125 126 Non-public functions (static ones, and those marked SRD_PRIV) don't need 127 to have @since markers. 128 129 The @since tag should be the last one, i.e. it should come after @param, 130 @return, @see, and so on. 131 132 133Protocol decoder guidelines 134--------------------------- 135 136 - The 'desc' metadata field for a protocol decoder, which contains a 137 short, one-line description of the protocol/bus, should be at most 55 138 characters long, and end with a full stop. This short description can be 139 displayed on the command-line using "sigrok-cli -V -l 3", or in various 140 different places in GUIs. 141 142 - Longer, multi-line descriptions should be placed in the protocol 143 decoder's __init__.py file as docstring. It can be viewed (for a specific 144 protocol decoder, e.g., UART) via "sigrok-cli -a uart", or in various 145 other places in GUIs. 146 147 - Generally use strings for states (of the PD state machine), not integers. 148 This avoids having to keep a list of state definitions at the top of file. 149 The performance overhead for this is negligible in practice. 150 151 Recommended: 152 self.state = 'IDLE' 153 self.state = 'GET STOP BIT' 154 Not recommended: 155 self.state = IDLE 156 self.state = GET_STOP_BIT 157 (where IDLE = 0 and GET_STOP_BIT = 1, for example) 158 159 - Generally use strings for commands/IDs in generated protocol packets. 160 This avoids having to know magic numbers of the PD in higher-level PDs. 161 The performance overhead for this is negligible in practice. 162 163 Recommended: 164 self.put(x, y, p, ['STOPBIT', 0, 0]) 165 self.put(x, y, p, ['ADDRESS READ', 0x51]) 166 Not recommended: 167 self.put(x, y, p, [STOPBIT, 0, 0]) 168 self.put(x, y, p, [ADDRESS_READ, 0x51]) 169 (with STOPBIT = 3 and ADDRESS_READ = 7, for example) 170 171 - Use ALL-CAPS names for PD states and protocol packet commands/ID. 172 Words should be separated by spaces (not underscores or the like). 173 174 Recommended: 175 'FIND ADDRESS', 'GET TEMPERATURE', 'START' 176 Not recommended: 177 'FIND_ADDRESS', 'Get Temperature', 'start' 178 179 180Testsuite 181--------- 182 183You can run the libsigrokdecode testsuite using: 184 185 $ make check 186 187 188Protocol decoder test framework 189------------------------------- 190 191Please see the sigrok-test repository for a protocol decoder test suite that 192checks the decoded data of various PDs against known-good reference data. 193 194 195Release engineering 196------------------- 197 198See 199 200 http://sigrok.org/wiki/Developers/Release_process 201 202for a list of items that need to be done when releasing a new tarball. 203 204