[03:43] tjaalton: Are input-device properties meant to actually work on Intrepid? I can see that you've patched xinput and libxi to support them, but xinput can't seem to talk properly. [03:45] (upstream xinput seems to want XI2 support to do anything with input properties... does the server only return them if it detects that the client knows about XI2?) [07:31] wgrant: it should work, the properties stuff has been backported to XI 1.5 [07:31] but I have to double-check it when I have more bandwith [07:34] our evdev doesn't support it yet, that's known [07:36] Ah. [07:37] And I guess that somebody will need to teach Synaptics about it. [07:37] I can see a 'Device Enabled' property, and that's all. [07:37] it should support it as well [07:37] Which doesn't appear in any of the listings that I've seen on the web. [07:38] I'll try my experimental evdev driver to make sure the plumbing is ok [07:41] yeah, evdev works so the synaptics driver is botched somehow [07:41] gone -> [07:41] See you. [07:42] Thanks. [10:53] wgrant: you could try patching out "#if GET_ABI_MAJOR(ABI_XINPUT_VERSION) >= 3", then it should work [10:54] from src/properties.c [11:26] wgrant: don't bother, I'll upload a new version [14:31] tjaalton: Boom. [14:31] tjaalton: dlopen: /usr/lib/xorg/modules/input//synaptics_drv.so: undefined symbol: SetProperty [14:42] great [14:44] Indeed. [14:44] So it does really need the new XI? [14:45] Or is just getting confused? [14:51] I think it's the latter [14:52] anway, I'll revert the patch for now [14:57] that's what you get without hardware to test on [14:57] although, in this case loading the driver would've been enough [15:00] I'll poke at it a bit and try to get it working. [15:03] for evdev to make it work with XI 1.5 the only thing it needed was to rip off the #ifdef's [15:04] Brb, restarting with a hopefully-fixed driver. [15:04] and Bool SetProperty is defined in synaptics.c, that's why I think it's a problem with the driver code [15:07] It's defined in properties.c, actually. [15:07] Which probably wasn't being linked in. [15:07] ok, got around upgrading my development Geode workstation to Intrepid and I confirm: odd artifacts where the cursor is, using -geode on X 1.5 [15:19] tjaalton: Right, got it. [15:19] properties.o wasn't being linked in. [15:20] Which probably means that src/Makefile.in wasn't being regenerated. [15:27] wgrant: yeah I saw the same.. [15:28] I'll fix the package, again, shortly [15:28] tjaalton: Running automake fixes it, and produces just a tiny diff. [15:28] And it works excellently. [15:28] Thanks. [15:29] wgrant: excellent :) [15:29] dinner-> [15:34] tjaalton: You wouldn't be able to upload the fix for bug #262292 as well, would you? [15:34] Launchpad bug 262292 in xfree86-driver-synaptics "[intrepid] clicking is not working after merging" [High,Triaged] https://launchpad.net/bugs/262292 [15:38] wgrant: yep [15:40] tjaalton: It is fixable through xinput now, but it's still annoying. [15:40] yes, it's been reverted upstream too [15:40] And it for some reason won't let me set the necessary property to have the right number of items. [15:43] What odd behaviour. [15:44] I'm trying to override the tap actions through xinput, but it will only accept 6 arguments for that property, when it needs (and has initially) 7. If I give it a different name, it will accept 7 fine. [15:45] Is that likely to be a driver or server bug? [15:45] I'm not sure, there are some limitations to the settings that can be changed atm [15:47] I can change it fine, just not to the right length. [15:47] No error is given, but it doesn't take effect, and watch-props doesn't see it. [15:48] maybe a driver bug [15:48] Ah. [15:49] It seems that anything that wants more than one will accept only one less than it should actually take. [15:49] * wgrant pokes in the code. [15:55] jcristau: does autoreconf generate config.h.in? [15:56] yeah, autoheader does, iirc [15:56] I'll autoreconfify synaptics, and noticed that the file is changed [15:57] * wgrant deletes lots of = [15:58] tjaalton: I've located and am fixing the off-by-one property issue, so xinput will be useful on it shortly. [15:58] Will you accept such a patch? [15:58] wgrant: excellent [15:58] sure, if it works properly after that. You should also send it to xorg@ :) [15:59] Of course, I shall. [15:59] Testing now. [16:03] Neeeed mooore baaandwidth. [16:04] I've got maybe 5kt/s [16:05] I've 64Kbps for another day or two. [16:05] kt? [16:05] oops [16:05] KB/s [16:05] Ow. [16:06] GPRS ftw [16:06] Sounds expensive. [16:07] But I guess you're not in .au. [16:07] nah, maybe 10EUR /month [16:07] 3G (~128Kbps) where there is coverage [16:08] built-in adapter in the laptop :P [16:08] 1.5c/KB on some of the more common plans here. [16:08] For both 3G and GPRS. [16:08] uh [16:09] It would go down a bit if you were to get a few hundred megabytes. But it's still very expensive. [16:10] OK, now I've tested this trivial patch, and it works well. [16:10] cool, post it somewhere [16:10] You need to delete a total of two characters. [16:10] Line 458, line 470. >= should be just >. [16:11] (src/properties.c) [16:13] if (prop->size >= MAX_CLICK and MAX_TAP? [16:13] Yep. [16:22] ok, the driver built fine, uploading [16:22] three uploads in one day [16:23] Thanks! [16:23] which is nice, Top10 here I come :) [16:23] Heh. [16:24] so, send the patch to xorg@lists.freedesktop.org [16:26] Any particular format/subject? [16:26] I guess I'll check the archives. [16:26] I don't think it matters much [16:31] 'git format-patch origin/master; git send-email' works [16:31] i think xorg@ is subscriber-only though [16:36] * wgrant is rejected, so subscribes. [16:49] * wgrant -> bed [16:49] Thanks for the uploads today, tjaalton. [16:49] wgrant: thanks for testing, and patches :)