| | <<<Back 1 day (to 2020/03/10) | Fwd 1 day (to 2020/03/12)>>> | 20200311 |
| avih | ator: i think you should document that js_Report should never throw, or even access J other than for js_getcontext. the mujs code is not ready for it throwing, and if used during panic then js_Report should probably not even use js_try | 14:28.34 |
| | (in fact, a more correct API would probably be js_Report(void *context, const char *message), but this ship has probably sailed by now | 14:30.01 |
| ator | avih: yeah... the idea is to use the js_getcontext if you need the user context | 14:31.17 |
| avih | sure, but it's not documented | 14:31.34 |
| ator | should probably have passed the uctx instead of the js_State, but as you said, that ship has sailed | 14:31.34 |
| | and won't prevent someone from doing weird stuff with the state anyway. will make a note in the docs! | 14:31.48 |
| avih | thx | 14:31.56 |
| | ator: would you consider supporting official js_utf8_push{string,lstring,literal} which convert the input to cesu8 if required, and similarly js_utf8_tostring (and possibly js_utf8_tolstring in the future) which returns it as utf8? i.e. just for the "pure" string insertions/fetch, while leaving all other APIs as is? | 15:37.11 |
| | these could implicitly provide conversion utils between utf8/cesu8 to the user: js_pushliteral(J, str); utf8 = js_utf8_tostring(J, -1) converts to utf8, and js_utf8_pushstring(J, str); cesu8 = js_tostring(J, -1) converts to cesu8. then the user can wrap the standard APIs with ones which convert as required (taking into account the additional stack entry used when using idx) | 15:41.01 |
| | at the second case actually js_utf8_pushliteral (which takes the literal if it doesn't need conversion, else pushes it as a cesu8 string). this way allocation is only used if a conversion is required | 15:43.46 |
| | the code change in mujs to support these would be rather minimal | 15:44.48 |
| | ator: "yes"/"no"/"i don't know yet" would suffice | 16:58.06 |
| ator | avih: yes. | 17:34.13 |
| avih | ator: cool. i'll send a PR on github at some stage then, hopefully within a week or two | 17:35.44 |
| | <<<Back 1 day (to 2020/03/10) | Forward 1 day (to 2020/03/12)>>> | |
Log of #mupdf at irc.freenode.net.