Development mailing list

I'm displeased if my message wasn't clear, i thought that i was expressing clearly the concept, i'll be more short and coincided next time going directly to the point. I wondered what you were talking about. But I guess you were just mixing up ports with application loopers (the latter get the broadcasts, and only accept messages). I was talking about looper being based on ports so there's potential of ...

On 03/31/2015 04:07 PM, Dario Casalinuovo wrote: The BMediaRoster::MessageReceived is actually rejecting anything which doesn't have the media_server and addon_server signature, but from a conceptual side why should we waste app_server and kernel resources if there's not any reason (and there will be not) to watch for all apps? Nice to read NULL ;-), what i understand is the problem does not exist, ...

Am 31.03.2015 um 16:14 schrieb Axel Dörfler axeld@xxxxxxxxxxxxxxxx: Am 30/03/2015 um 16:45 schrieb Dario Casalinuovo: Hello, i'm referring to the changes i proposed in #11922, is there any reason (that i can't see) for having the roster send obligatorily notifications for all apps running? I wondered what you were talking about. But I guess you were just mixing up ports with application loopers (the latter get the broadcasts, and only ...

Am 30/03/2015 um 16:45 schrieb Dario Casalinuovo: Hello, i'm referring to the changes i proposed in #11922, is there any reason (that i can't see) for having the roster send obligatorily notifications for all apps running? I wondered what you were talking about. But I guess you were just mixing up ports with application loopers (the latter get the broadcasts, and ...

The BMediaRoster::MessageReceived is actually rejecting anything which doesn't have the media_server and addon_server signature, but from a conceptual side why should we waste app_server and kernel resources if there's not any reason (and there will be not) to watch for all apps? Nice to read NULL ;-), what i understand is the problem does not exist, nevermind. ...

Hello, i'm referring to the changes i proposed in #11922, is there any reason (that i can't see) for having the roster send obligatorily notifications for all apps running? In the first place it looked to me a design flaw, and really i can't understand why it wasn't foregone, also fastly looking at registrar it seems shouldn't be a problem to extend the Watcher to have a list of teams and check if the BMessenger is interested to all teams or only to some. Yes it's a very little bit of computational cost added, since we should check this for every Watcher, but in the average case this isn't really relevant asymptotically. Instead there's the cost of having all those ports and ...

On 3/29/2015 13:15, Alexander G. M. Smith wrote: Looks correct to me. I see you used your exclude region feature to cut out the test window itself, otherwise you'd see the recursive out to infinity view of the whole screen (which happens if you run the VNC client on the same machine as the server). Yes, though the update events were still being fired for that occurrence, so I'm remedying that with a bit of a redesign. ...

looncraz looncraz@xxxxxxxxxxx wrote on Sat, 28 Mar 2015 16:37:42 -0500: Just 10MB, a quick video of it in action ;-) Looks correct to me. I see you used your exclude region feature to cut out the test window itself, otherwise you'd see the recursive out to infinity view of the whole screen (which happens if you run the VNC client on the same machine as the server). You can set a frame rate limit, and it only captures when you Update(). ...

On 2015-03-26 08:22, Dane Scott - TuneTracker Systems wrote: $200 so far, and we'll keep plugging it. If anybody's interested in working on this (or IS working on it), let me know. [1] Dane ...

On 3/28/2015 15:02, Alexander G. M. Smith wrote: looncraz looncraz@xxxxxxxxxxx wrote on Fri, 27 Mar 2015 12:10:21 -0500: Do you try to calculate the changes in the screen, or you just send the whole thing as fast as you can? There's a dirty rectangle algorithm in the VNC library code that compares rectangular areas from the current bitmap with the previously transmitted ...

looncraz looncraz@xxxxxxxxxxx wrote on Fri, 27 Mar 2015 12:10:21 -0500: Do you try to calculate the changes in the screen, or you just send the whole thing as fast as you can? There's a dirty rectangle algorithm in the VNC library code that compares rectangular areas from the current bitmap with the previously transmitted data and only sends the changed pixels. However, reading the whole screen to find those dirty areas is slow on some video boards (it's actually done in smaller segments to improve response time), and even slower with the BScreen copy-the-whole-screen-to-a-bitmap technique. ...

On Fri, Mar 27, 2015 at 12:10:21PM -0500, looncraz wrote: Are you are suggesting setting up a virtual video board in effect? That could be an interesting way of doing it, which would also speed up screen reading since it would reside in main RAM, not the occasionally much slower video memory. On the other hand, it would be simpler for the user to understand a program that merely shows what's on the real screen, remotely. A virtual video card (RemoteHWInterface) already exists for something like this, though I don't know the details. It should be immensely useful for ...

On Fri, 27 Mar 2015 11:49:40 -0600 Daniel headbulb@xxxxxxxxx wrote: I've been trying to build the VirtualBox additions but the compile fails. Gcc complains about -fpic . My next step is to run though the commands manually. My 2p worth: as a prelink user on GNU/Linux, I observe that Virtualbox links with at least some non-PIC libraries. ...

Wasn't something like this needed in the app_server for the VirtualBox video driver? I've been trying to build the VirtualBox additions but the compile fails. Gcc complains about -fpic . My next step is to run though the commands manually. Has anyone gotten the additions to build? On Fri, Mar 27, 2015 at 11:10 AM, looncraz looncraz@xxxxxxxxxxx wrote: ...

On 3/27/2015 09:24, Alexander G. M. Smith wrote: Currently VNCServer just screen (using a BDirectWindow for direct video memory access or alternatively BScreen-ReadBitmap()) scrapes the active workspace, and handles switching between different screen sizes and depths fairly well. Do you try to calculate the changes in the screen, or you just send the whole thing as fast as you can? ...

Hi all, I have done a bit of requirements elicitation [1] on the trac wiki, namely on ARM microarchitectures. Or in other words, I am trying to gather information such as developer ownership of ARM development hardware, so we can work out which devices we should try to target. It will also be helpful for people who are looking to buy an ARM board, but want to get one that is more common among Haiku developers in the hope that it will be supported sooner. ...

looncraz looncraz@xxxxxxxxxxx wrote on Thu, 26 Mar 2015 10:46:12 -0500: Yes, both are doable. I'm not certain as to how VNC remote viewing is handled, but I imagine it uses the RemoteHWInterface in the app_server, which will not enable the hardware cursor, so will always use a software cursor. Currently VNCServer just screen (using a BDirectWindow for direct video memory access or alternatively BScreen-ReadBitmap()) scrapes the active workspace, and handles switching between different screen sizes and depths fairly well. Are you are suggesting setting up a virtual video board in ...

On 3/26/2015 10:10, Alexander G. M. Smith wrote: Which reminds me to ask if it would be easy to add a way of getting the current cursor shape, for use in VNC remote viewing of the screen. Or a way to force it to use the software cursor. It would also be nice to have a list of the dirty rectangles for the whole screen. Doable? - Alex Yes, both are doable. I'm not certain as to how VNC remote viewing is ...

In [haiku-commits] Re: BRANCH looncraz-github.CAP-dirty [f8c99665426f] on Wed, 25 Mar 2015 11:04:55 -0500, looncraz looncraz@xxxxxxxxxxx wrote: And, on to the subject of the mouse cursor, if no hardware cursor is enabled, and the cursor is above a BDirectWindow we have an opportunity with the composite control loop to update it just a few microseconds before the refresh. Of course, all this assumes our refresh timing is accurate enough. Which reminds me to ask if it would be easy to add a way of getting the current cursor shape, for use in VNC remote viewing of the screen. Or a ...

On Thu, Mar 26, 2015 at 8:22 AM, Dane Scott - TuneTracker Systems tunetrackersystems@xxxxxxxxx wrote: $200 so far, and we'll keep plugging it. If anybody's interested in working on this (or IS working on it), let me know. Dario (aka Barrett) was working on it: ...