| <<<Back 1 day (to 2020/10/08) | Fwd 1 day (to 2020/10/10)>>> | 20201009 |
malc_ | Robin_Watts: 32e4e8b4bcbacbf92af7c88337efae21986d9603 is weird, ptrdiff_t is a signed type and you pass it as a size_t argument to memset, in your eyes it's an improvement because size_t can be 64bit? what about LLP (or whatever the thing that amd64 windows uses) | 17:28.59 |
Robin_Watts | w, h, n are signed. | 17:29.48 |
| stride is signed. | 17:29.59 |
| the important thing is that sizeof(pdfdiff_t) == sizeof(size_t) | 17:30.14 |
malc_ | huh? | 17:30.31 |
| sez hu? | 17:30.35 |
Robin_Watts | ptrdiff_t has to be large enough to cope with the difference between any two pointers. | 17:31.19 |
| So on a 64bit system, it'll be 64bits. | 17:31.37 |
malc_ | yes | 17:31.52 |
| but size_t? | 17:31.55 |
Robin_Watts | w*h*n is only ever 32bits on a system where int is 32bits. | 17:32.12 |
| ((ptrdiff_t)w) * h * n will be 64bits on a 64bit system, even if int is 32bits. | 17:32.39 |
segfault57 | I'm just looking at mujs, it seems a very nice system. But the documentation is lacking, I was wondering if it's possible to load a json string as an object. Any hints would be highly appreciated. | 17:32.52 |
Robin_Watts | segfault57: You need to talk to ator, I suspect. He may be having dinner at the mo cos he's in sweden. | 17:33.31 |
malc_ | Robin_Watts: this still does not answer my size_t objection | 17:33.37 |
Robin_Watts | malc_: You think it's likely that a size_t will be a different size to ptrdiff_t ? | 17:34.04 |
malc_ | Robin_Watts: sure | 17:34.12 |
segfault57 | size_t is supposed to be the same size as the size of pointer. | 17:34.17 |
Robin_Watts | malc_: Give me an example of such a system. | 17:34.31 |
segfault57 | size_t and ptrdiff_t should be the same size. | 17:35.03 |
malc_ | Robin_Watts: huge model on DOS compilers circa ~90 | 17:35.08 |
Robin_Watts | malc_: I care not about such things. | 17:35.21 |
malc_ | segfault57: can you, please, substantiate that assertion with C&V from the standard | 17:35.34 |
segfault57 | what is C&V? | 17:35.54 |
malc_ | Robin_Watts: point being - it is possible | 17:35.57 |
| segfault57: chapter and verse | 17:36.09 |
Robin_Watts | malc_: On such a system, what would ptrdiff_t and size_t be ? | 17:36.09 |
segfault57 | malc_, you got me. I only work on Linux. | 17:37.33 |
malc_ | Robin_Watts: you mean huge model dos? | 17:39.17 |
Robin_Watts | malc_: Yes. | 17:40.11 |
segfault57 | @actor, I need to load a string as a JSON object, is it possible? my current workaround is to create a temp variable and prepend it to the json string like this "var tmp = { ...}", load this string, and use "tmp". But this is not ideal. | 17:40.40 |
malc_ | Robin_Watts: https://www.youtube.com/watch?v=5nKxmoP8354 | 17:40.50 |
Robin_Watts | I think we probably fall into a massive heap if run on a segmented memory model anyway. | 17:41.00 |
| so I refuse to worry about such things. | 17:41.14 |
malc_ | Robin_Watts: comment would be nice though | 17:41.46 |
Robin_Watts | I could have used ssize_t, but I fear that's not always available. | 17:42.19 |
malc_ | Robin_Watts: yes, it's posix, not C. https://stackoverflow.com/questions/55190317/where-is-ssize-t-defined-in-linux | 17:46.18 |
Robin_Watts | hence ptrdiff_t was the best I could come up with. | 17:46.51 |
malc_ | i still think comment would be nice | 17:48.56 |
| Robin_Watts: https://devblogs.microsoft.com/oldnewthing/20200728-00/?p=104012 | 17:56.34 |
Robin_Watts | yes, I looked at that, and went screaming away as I always do when confronted with segmented models. | 17:57.15 |
malc_ | i always used tiny model, and then wrote my own dose extender to not deal with it on 32bit platforms" | 18:02.07 |
| <<<Back 1 day (to 2020/10/08) | Forward 1 day (to 2020/10/10)>>> | |