| <<<Back 1 day (to 2014/10/21) | 20141022 |
henrychen | dear all, how can I improve the performance of 'MuPDFCore_saveInternal' ? It spends lots of time to save the file . | 02:17.39 |
kens | Great, so we actually already have a convenient place to do device-specific initialisationof data, but its less than obvious. The easy place to do it is in the finish_copydevice method. | 08:26.24 |
| We can't do an 'init' routine in gx_device_init, because the device procs aren't set up at that time (that is, any procs which are 'default' aren't set up, they are still NULL). Those don't get set up until we do a setdevice. The initialisation of devices is totally insane. | 08:27.45 |
| Goodness knows when devices get set up by non-PostScript interpreters, I guess I'll have to find out :-( | 08:28.11 |
rayjj | kens: I saw the comment about initialization. I knew that device procs weren't set yet -- fill_in_procs is one of the things that I would think that a device specific init would do. | 15:34.39 |
kens | rayjj the vcei procs is a different problem. | 15:34.59 |
| device procs* | 15:35.05 |
rayjj | but it sounds like you have found a method | 15:35.13 |
kens | I opted (so far) not to have a init method, turns out that we can initialise device specific data in the finish_copydevice | 15:35.41 |
rayjj | kens: right, that's what I meant by "you found a method" (I saw that in the logs) | 15:37.31 |
kens | Yes, it seems ot be reliable enough, it look slike that's pretty much what the proc is intended for | 15:38.00 |
rayjj | not such an obvious name :-) | 15:38.25 |
kens | Now I'm back to my previous investigation, with support emails for light relief | 15:38.30 |
| Agggh !!! Yet another 'special case' :-( | 15:39.31 |
| Initialisation of parts of devices gets done in *so* many places | 15:40.40 |
| Hmm, so WIndows 10 is going to allow me to use fingerprints or my phone as an authentication device. What if I don't have a fingerprint scanner or a smart phone ? | 15:48.58 |
Robin_Watts | kens: You'll need to purchase a mMicrosoftFinger(TM) | 15:50.24 |
kens | Well from the rest of the article, it does look like they are going to use security as a reason to force people to use the Windows store | 15:51.06 |
| SO you may be correct | 15:51.14 |
Robin_Watts | Fingerprints can only be registered to a single account. If you want to switch account then you'll have to go see a microsoft certified plastic surgeon. | 15:51.14 |
kens | That's OK< its MS implementing it, I'll just use someone else's fingers, they'll never know | 15:51.39 |
Robin_Watts | kens: but Microsofts fingers won't be entirely compatible with the international standard... | 15:52.27 |
kens | Robin_Watts : they'll be compatible in a 'future' version | 15:52.48 |
| After which MS Fingers will become the defacto standard | 15:53.07 |
Robin_Watts | Apple will invent a hand with just a single finger and claim it's more usable. | 15:53.18 |
kens | Or at least more aesthetically pleasing | 15:53.32 |
| Of course Linux will aloow you to use the appendage of your choice | 15:54.04 |
rayjj | Yeah, and Apple can hold up that single finger to Microsoft (we all know which one) | 15:54.10 |
Robin_Watts | kens: The Linux power users will have extra fingers grafted on... | 15:54.30 |
kens | You mean they don't already ? | 15:54.39 |
rayjj | Robin_Watts: they already do, so they can type faster :-) | 15:54.53 |
| and VMware will have a product that lets you slip on alternate fingers | 15:56.33 |
| emacs users will have to scan 4 fingers at once | 15:57.26 |
kens | heads off for the night. GOodnight folks | 16:37.28 |
| Forward 1 day (to 2014/10/23)>>> | |