| <<<Back 1 day (to 2015/10/05) | 20151006 |
tor8 | Robin_Watts: commits on tor/master for review, and tor/glut for consideration to merge to master | 09:21.55 |
Robin_Watts | tor8: OK. Will try to look today. | 09:22.16 |
| tor8: The mupdf master commits (4 of them) look fine to me. | 13:54.00 |
| I'd OK'd the previous 2 the other day, I think. | 13:54.46 |
kens | Hmm something wrong with one of Marcos' scripts I think, I just got 8 emails about the Luratech regression report | 14:03.09 |
Robin_Watts | tor8: Your prev_char helper function can overrun the start of a string if it's fed a malformed UTF8 string. | 14:15.12 |
henrys | URW has made the schedule 13 week nows so December for the fonts. I suggested GG hire URW another designer, but I don't think that's likely | 14:15.13 |
Robin_Watts | tor8: I think you need prev_char(p, start); | 14:16.15 |
tor8 | Robin_Watts: yeah, that's probably a good idea | 14:16.36 |
| once we start using it for editing forms data, when we can't be sure of the correctness of input data | 14:17.03 |
Robin_Watts | toggle_fullscreen has the code therein #if 0#'d out. | 14:18.26 |
tor8 | yeah, I haven't found an equivalent GLFW function :( | 14:18.57 |
Robin_Watts | My brain can't cope with all that at once, but 1) it doesn't break any of the existing stuff, and 2) I can't see anything horrible in there, so I say go for it. | 14:20.37 |
tor8 | Robin_Watts: three new commits at the tip of tor/glut. an epub fix, updating the freetype submodule to 2.6.1 and a fix for the potential overrun you just spotted. | 14:23.34 |
Robin_Watts | tor8: As a stylistic thing... I dislike functions that do: int some_fn(arg) { if (arg) { ..... } return x; } | 14:25.23 |
| I'd far rather see int some_fn(arg) { if (!arg) return NULL; .... } | 14:26.16 |
| It reads nicer to me, and avoids rightwards drift. | 14:26.47 |
| Otherwise, all look good to me. | 14:27.26 |
henrys | marcosw: hold on... we want that open 696203 as an enhancement maybe. | 14:28.31 |
marcosw | you mean 208 | 14:28.51 |
henrys | oh I see you just verified the dup | 14:29.07 |
marcosw | was just writing you an email asking if 208 should be changed to an enhancement | 14:29.14 |
henrys | marcosw: yes it should | 14:29.37 |
marcosw | done | 14:29.49 |
henrys | marcosw: btw the coverage links are 404'd on the dashboard. I do actually look at them. | 14:30.35 |
| kens: that FirstPage regression has got me a little nervous. | 14:31.10 |
kens | I haven't had a chance to look at it yet, I'm answering the earlier one | 14:31.31 |
| If its the same person it may well be nothign to worry about | 14:31.43 |
marcosw | henrys: oops, not sure why that should be. I'll check. | 14:32.01 |
kens | henrys I did test the ijs device, and it worked for me, though obviously thre's nothign I can send it to | 14:33.59 |
henrys | mvrhel_laptop: I'm sort of mixed on whether we should have ray start profiling on a Pi or somethnig similar. I don't have a feel for how "real" this arrangement is. As I said in email ... | 14:34.10 |
kens | As for X11 I am limited in my ability to test that | 14:34.11 |
| However, I don't thnk ths can be considered a regression, since it apparently works perfectly well if you don;t use the new feature. | 14:34.39 |
Robin_Watts | henrys: We shouldn't do *any* work on stuff specifically for that company. | 14:34.48 |
| They have a history of eating our time for no return. | 14:35.00 |
tor8 | Robin_Watts: thanks. | 14:35.17 |
Robin_Watts | Doing general stuff (such as checking for hotspots etc) doesn't seem unreasonable though. | 14:35.32 |
| (IMHO etc) | 14:35.57 |
henrys | agreed, I'll talk to ray when he comes around. | 14:36.10 |
| how's the mupdf release? | 14:36.17 |
| we've got to get out of double digit bugs or Miles will call me ;-) | 14:37.14 |
Robin_Watts | From my point of view, there are a couple of bugs that I'd like to see fixed before release. (696241 being one). | 14:37.37 |
| henrys: We're in triple digits. You're fine. | 14:37.47 |
henrys | actually I have a trivial one to fix which will bring us down to 11, so just 2 more... | 14:38.44 |
Robin_Watts | oh, I thought you meant mupdf ones :) | 14:38.59 |
mvrhel_laptop | henrys: just sent a reply to your email. I agree I don't want to spend much time on it, but I think we do have an opportunity to get some comparison between us and GG | 14:39.01 |
henrys | customer bugs | 14:39.14 |
Robin_Watts | mvrhel_laptop: If we can get information from MQ, yes, absolutely. | 14:39.33 |
henrys | mvrhel_laptop: yeah we were just discussing that. | 14:39.34 |
mvrhel_laptop | Robin_Watts: that is the idea | 14:39.51 |
henrys | chrisl: I don't know if you saw about URW's new schedule is for December | 14:40.10 |
| s/about/above | 14:40.20 |
mvrhel_laptop | once we see how we compare on his system, then we can verify that we are getting the same time that he is seeing that if we want we can profile on the ARM | 14:40.29 |
chrisl | henrys: I did, yes. Not sure what to make of it..... | 14:40.34 |
mvrhel_laptop | he said it is straight forward with there tool chain | 14:40.42 |
| s/there/their/ | 14:40.50 |
Robin_Watts | mvrhel_laptop: Right. But doing anything much before we get to that stage seems like a waste of cycles to me. | 14:41.12 |
chrisl | Hmm, not much point in doing a clusterpush...... | 14:41.27 |
mvrhel_laptop | Robin_Watts: yes I agree. there is not much to do until we see if there is a diff and that it is real | 14:41.45 |
| if we do find out that there is a real difference and its of the magnitude that he is talking about we absolutely need to understand what is going on though | 14:42.17 |
| if we have any interest in getting into the japanese market | 14:42.30 |
chrisl | That may, of course, not be possible unless we can also see the GG output | 14:42.57 |
mvrhel_laptop | well yes see my email | 14:43.12 |
Robin_Watts | mvrhel_laptop: Agreed. | 14:43.14 |
mvrhel_laptop | chrisl: he will be running gs and gg side by side | 14:43.27 |
| and should be set up to print I am certain | 14:43.38 |
Robin_Watts | If you can get to that stage, it will be a brilliant step forward. | 14:43.51 |
mvrhel_laptop | I think its the only way we can make any progress | 14:44.16 |
| MQ even thinks so. He does not like just passing numbers along from us | 14:44.31 |
henrys | chrisl, kens does harlequin have a sophisticated display list architecture that could say detect blank regions and that sort of thing? There's huge gains to be had from that with rops. | 14:44.48 |
kens | 'sophisticated' is a matter of opinion. Its certainly complicated | 14:45.14 |
chrisl | More like baffling...... | 14:46.12 |
marcosw | chrisl: clusterpush's get priority over commits, so you'd go to the top of the queue. | 14:46.33 |
chrisl | marcosw: I know, I was just being facetious | 14:46.52 |
marcosw | chrisl: thought maybe you were looking for an excuse to quit early :-) | 14:47.16 |
henrys | there isn't much we can do about "fixing" anytime soon in gs, that would be a big project | 14:47.22 |
Robin_Watts | henrys: I dunno.... | 14:47.38 |
henrys | possibly kens device architecture could be used for that purpose. | 14:47.46 |
kens | eh, what purpose ? | 14:47.57 |
chrisl | Fixing what in gs? | 14:48.27 |
henrys | I'm thinking about recording regions that have been marked so we can possibly avoid some rops | 14:48.29 |
kens | I'd have thought that was better done in the clist | 14:48.45 |
chrisl | We already do that for transparency in the clist | 14:48.51 |
henrys | but I'm getting ahead of myself. Let's wait for the data | 14:48.52 |
Robin_Watts | We could have a pass through device that maintained a 'dirty grid' for the screen. put that in front of any clist writing step. Then when we do getbits, if the area is in a clear bit of dirty grid, we can trivially return 'clear'. | 14:48.59 |
| s/screen/page/. | 14:49.13 |
chrisl | I assume raster ops are handled after the clist? | 14:49.25 |
kens | Surely they must be performed at renrdering time | 14:49.41 |
Robin_Watts | In fact, the clist doesn't come into it. | 14:49.50 |
chrisl | One would hope.... | 14:49.57 |
Robin_Watts | It just needs to be before the renderer. | 14:49.58 |
| kens, absolutely, brain fart, sorry. | 14:50.16 |
chrisl | As I said, we already have a record of bands that require transparency, so possibly that could be extended for this purpose | 14:51.16 |
henrys | another idea I have that I think would work is doing rops in source space, not devices space. There could be huge savings in that, but it's not exactly correct | 14:51.24 |
kens | I'm not sure why that helps, but thenm I don;t understand ROPs so..... | 14:51.54 |
henrys | chrisl: there is some optimization already. | 14:51.54 |
mvrhel_laptop | henrys: does the fact that GG shows no timing diffs when going out to difference device types tell you anything? | 14:52.23 |
| s/difference/different | 14:52.33 |
chrisl | How accurate was the timing? | 14:52.52 |
mvrhel_laptop | well you should see our differences..... | 14:53.06 |
henrys | mvrhel_laptop: sound like voodoo to me. How can that possibly be? | 14:53.12 |
mvrhel_laptop | I agree | 14:53.15 |
| it seemed very odd. MQ thought it was odd too | 14:53.28 |
chrisl | I was thinking if they always rendered to RGB and then converted to the required output colors...... | 14:53.49 |
henrys | mvrhel_laptop: yeah I'll send you some interesting rop tests today | 14:53.52 |
mvrhel_laptop | henrys ok thanks | 14:54.00 |
| chrisl: that was my thinking too | 14:54.12 |
henrys | anything else meeting wise. A good meeting for ray to be at but he'll catch up reading the logs. | 14:54.13 |
Robin_Watts | It's not that MQ is timing the GG interpret->display list stage, and not the displaylist -> output representation stage? | 14:54.16 |
chrisl | mvrhel_laptop: in which case I'd have expected a measurable difference between the contone and halftoned output, and/or RGB and CMYK output.... | 14:54.48 |
Robin_Watts | Dunno if the GG display list is device independent or not (MuPDF's is, gs's isn't) | 14:55.07 |
kens | I'm fairly sure it can't be | 14:55.27 |
| Unless they changed it *very* significantly, which I would have thought was unlikely | 14:55.50 |
mvrhel_laptop | Robin_Watts: he is timing it in the same way as we are doing gs. Which is to start with the interpreter | 14:55.55 |
Robin_Watts | ok. | 14:56.11 |
chrisl | There is a chance that HQN detects discarding the output, and doesn't render when doing so | 14:56.30 |
kens | Ah, a 'cheat' device :-) | 14:57.08 |
mvrhel_laptop | anyway, it will be a few weeks until he is back from his travels and I will see about getting back together with him to see where we are. so nothing really to do until after that | 14:57.09 |
chrisl | kens: well, we had that option in Jaws..... | 14:57.24 |
henrys | kens: frequently images used for rasterops are 300 dpi in color pcl so why not do the rops at that resolution instead of device 600 or 1200, quite a good speedup where each pixel is visited | 14:57.27 |
chrisl | henrys: don't you have a problem getting the backdrop then? | 14:58.17 |
henrys | chrisl: yup doesn't work for destination operations, it works for patterns and images which are quite common | 14:58.55 |
Robin_Watts | henrys: So what happens when I put a 300dpi image rop'd over the top of a 200dpi one ? | 14:58.58 |
mvrhel_laptop | that is a cheat | 14:59.13 |
Robin_Watts | as an optimisation for rops that don't involve 'd' that seems reasonable. | 14:59.47 |
kens | From one of our bug reports, it seems some 'PCL' devices simply don't implement at least *some* rops at all. | 14:59.49 |
marcosw | HQN not rendering when the output is discarded sounds like something VW would have thought of. | 14:59.54 |
Robin_Watts | That would be simple to test; just get MQ to redo timings writing to a file. | 15:00.46 |
henrys | off to skype now... | 15:00.54 |
chrisl | I wonder if GG actually have any PCL customers...... | 15:01.04 |
henrys | chrisl: Marvell? | 15:01.17 |
chrisl | henrys: do we know they've actually shipped anything? | 15:01.38 |
henrys | no I don't maybe mvrhel_laptop knows | 15:02.35 |
| oh lord I have 75 emails | 15:03.03 |
kens | all TOr's commits | 15:03.14 |
henrys | ahh | 15:03.17 |
kens | and cluster resutls | 15:03.27 |
| there are a few real ones in there | 15:03.34 |
mvrhel_laptop | I know company M is shipping lots of parts. I believe GG is what they offer if the customer does not have their own | 15:04.17 |
chrisl | marcosw: still here? | 15:09.17 |
marcosw | chrisl: yes | 15:10.30 |
chrisl | marcosw: I just realised that my last clusterpush failed every gs on peeves with the "cannot find gs_init,ps" error, but I'd already pushed a bmpcmp test :-( | 15:11.09 |
marcosw | chrisl: amazingly enough I just had the cannot find gs_init.ps error on my local i7, which as far as I know has never happened before. | 15:12.35 |
kens | It seems to be getting more frequent | 15:12.59 |
chrisl | It's very odd that it seems mostly on peeves | 15:13.44 |
marcosw | this was unrelated to the cluster, just compiling head via make -j99 | 15:13.50 |
| chrisl: wonder if it's a makefile issue with parallel builds... | 15:14.03 |
| those tend to be system dependent. | 15:14.17 |
chrisl | marcosw: it's possible - I still expect to see it more often on the cluster, then | 15:14.46 |
marcosw | i'll setup a local test on peeves to see if I can reproduce it by doing repeated build/test cycles (keeping the make output). | 15:16.18 |
| my local problem went away when I did a make -j1. Then after a make clean a make -j99 worked as well (-j99 is what i use a default). | 15:17.09 |
chrisl | marcosw: If it is that, it will be a b*tch to track down :-( | 15:17.42 |
marcosw | i was thinking that there might be something in the make output that would indicate the cause. being able to reproduce it even randomly without having to run clusterpush would make me happy. | 15:19.48 |
chrisl | marcosw: I guess I could throw together a script to test it here..... | 15:20.40 |
| marcosw: I assume we don't keep the make output for a successful build? | 15:21.15 |
marcosw | On the cluster? yes, until the next cluster run. | 15:21.58 |
chrisl | Bum - I didn't realise I'd got errors, 'cause there were a number of diffs, too | 15:22.40 |
marcosw | chrisl: peeves is the only system that I've seen it fail on (other than the one time on my i7), so I expect it's going to be hard to reproduce anywhere else. | 15:23.41 |
chrisl | marcosw: yes :-( | 15:23.55 |
henrys | chrisl: have you run into Xorg hogging cpu time, I'm on 15.04. It started for me with some update, I've read about it happening to others with different configuration nothing suggested is working for me. | 16:18.09 |
chrisl | henrys: I have yes, but it's always settled down after a couple of minutes, so I haven't worried about it - so far..... but I only updated to 15.04 last night | 16:19.15 |
henrys | chrisl: hits me periodically sometimes settles down sometime not without logging off. | 16:21.51 |
marcosw_ | chrisl: just reproduced the âCan't find initialization file gs_init.ps.â issue on my i7. Only took 19 makes. | 16:21.58 |
| looking at the make output now. | 16:22.10 |
chrisl | henrys: I'll keep an eye on it now that I've updated.... of course, it could depend on the graphics card in use | 16:22.35 |
henrys | fredross-perry: did you have anything for the skype SOT meeting? Are you all set up to develop? | 16:24.23 |
fredross-perry | I am all set up, yes. Thanks. | 16:24.38 |
marcosw_ | the build that results in a broken gs is missing a lot of âincluding: './lib/../Resource/Init/gs_fntem.psââ statements. so that make sense. | 16:24.56 |
| but does include âCannot open gconfig file ./obj/iconfig.hâ. Does that give you enough to fix it? | 16:25.19 |
chrisl | Erm, possibly..... | 16:28.43 |
| marcosw_: can you mail me the entire make output? | 16:29.44 |
marcosw_ | sure. | 16:29.48 |
| email sent. includes both failed and good output. | 16:31.04 |
chrisl | Thanks - I *think* I see what the trouble might be..... | 16:31.22 |
marcosw_ | iâm surprised that it was so easy to find on my i7. A 5% failure rate should have shown up before (otoh, maybe itâs a lot less than 5% and I just got really luckly). | 16:31.56 |
Robin_Watts | henrys: So this new PC of yours... water cooled ? | 16:33.18 |
chrisl | marcosw_: Ah, there is a missing dependency, but I'd have expected that to result in a failed build, rather than what we're seeing..... | 16:33.25 |
Robin_Watts | saw a water cooled laptop in a magazine earlier. Bonkers. | 16:33.59 |
marcosw_ | it does seem odd that mkromfs doesnât return an error code⦠| 16:34.32 |
Robin_Watts | http://www.pcworld.com/article/2981240/software-games/up-close-with-the-asus-rog-gx700-a-massive-watercooled-gaming-notebook.html | 16:35.08 |
henrys | https://www.pugetsystems.com/serenity.php | 16:35.17 |
| I really liked working with those guys, it cost a little extra but well worth it. | 16:35.47 |
| there up by mvrhel_laptop | 16:36.03 |
| s/there/they are located | 16:37.13 |
marcosw_ | henrys: wow, they go up to a 95W processor. no fans at all? | 16:37.43 |
Robin_Watts | They must have fans, just quiet ones. | 16:38.21 |
henrys | there are fans - see fan tuning in the link I sent | 16:38.40 |
| kind of cool to see the thermal imaging and all before they ship the system. | 16:39.42 |
marcosw_ | does the fan tuning software work with linux? a lot of the high-end fans Iâve bought come with windows only software (controlled through usb) | 16:41.12 |
Robin_Watts | The people building mine don't do anything that smart. In fact they don't even appear to be answering their sodding email. | 16:41.35 |
henrys | marcosw: yeah it was tested and benchmarked with kubuntu which is what I ordered | 16:41.55 |
| the service has been absolutely phenomenal | 16:42.10 |
Robin_Watts | But I specced a case with silent fans, and a mahoosive silent CPU cooler, and a GPU with silent (in 2d mode) fans. | 16:42.31 |
| and a silent PSU. | 16:42.48 |
henrys | With these folks you spec out your own system but you also tell them what exactly you want to do and they will step in if they see something better for you. | 16:44.22 |
marcosw_ | My Amazon Echo just said âWhat is spidermanâs ideal job? A web designer.â. This was completely unprompted, Jill and I were having a conversation about image resizing and it presumably thought we were talking to it, otoh, the word spiderman was never said. | 16:44.26 |
henrys | marcosw_: I've heard good and horrible things about that contraption, how do you like it? | 16:45.49 |
marcosw_ | I think the Echo just randomly says stuff occasionaly if you havenât been using it to remind you itâs still there. | 16:46.06 |
| henrys: Iâve had mine for a long time and still donât know what use it is. We mainly use it for getting weather reports and the current time. | 16:46.40 |
| If you ask it âAlexa, are you better than siri?â it says, âNot better, just differentâ. WHich is pretty much the exact same thing siri says. | 16:47.40 |
| you can stuff like âalex, play classical musicâ and it does. | 16:48.03 |
Robin_Watts | Speccing as close to the same machine as mine as possible on that, comes out 10% more expensive. That's not a huge amount to pay for service. | 16:51.25 |
marcosw_ | I only paid $99 for mine as an invited pre-order. I think if I had paid $179 I would have returned it. | 16:51.42 |
mvrhel_laptop | that looks cool | 16:55.16 |
Robin_Watts | Gotta get one of those. At least then there'd be one thing in this house that listens to what I say. | 16:56.07 |
henrys | Robin_Watts: yeah I thought my price came out reasonable given the service. And they really didn't try to see me anything, it was more the other they wanted to remove stuff I didn't need. | 16:56.39 |
| s/see/sell | 16:57.17 |
Robin_Watts | oops. I just found the same CPU I've got, and the price jumped a lot :) | 16:58.08 |
| Interesting that the CPU cooler they would put is in the smaller brother of the one I got. | 16:58.35 |
| Plus, importing it to the UK would be a killer. | 16:58.47 |
henrys | Robin_Watts: did you go to lenovo? | 16:58.55 |
Robin_Watts | No. Custom build from scan.co.uk | 16:59.08 |
chrisl | marcosw_: there's a commit in my repo which I think should solve the bulid problem - could you try it out, please? | 17:01.06 |
marcosw_ | chrisl: looks like the gs_init.ps failure occurs about 1 time in 40 on both my i7 and peeves. so iâm very surprised i have I havenât seen it more often on the cluster since we run 30 builds for every cluster run. Must be very dependent on cpu speed. | 17:01.26 |
| chrisl: will do. | 17:01.31 |
henrys | Robin_Watts: what CPU did you get? | 17:01.59 |
chrisl | marcosw_: thanks - I'll have to be heading off shortly | 17:02.15 |
Robin_Watts | Intel Core i7 5930K | 17:02.53 |
marcosw_ | chrisl: Iâll send you an email after enough tests after running enough tests to have a 95% confidence that itsâ fixed (guess my PhD is useful after all). | 17:03.10 |
Robin_Watts | (6 Core with Hyperthreading, 3.5GHz, 40 PCI-e lanes) | 17:03.15 |
chrisl | Thanks | 17:03.20 |
| Does the cluster filtering take wildcards? | 17:04.03 |
| Robin_Watts: ^^ | 17:04.08 |
Robin_Watts | chrisl: Not really. | 17:04.29 |
| It just looks for substrings. | 17:04.40 |
| what were you looking for ? | 17:04.44 |
chrisl | I was hoping to to "ppmraw.?00.0" so I got both Ghostcript and PCL | 17:05.14 |
Robin_Watts | -filter=ppmraw.300.0 -filter=ppmraw.200.0 | 17:05.32 |
| filters are 'or'ed together. | 17:05.47 |
chrisl | Oh, cool - thanks | 17:05.53 |
Robin_Watts | np. | 17:06.09 |
marcosw_ | chrisl: can you check the commit? I get an error: | 17:13.14 |
| make: *** No rule to make target 'obj/iconfig.h', needed by 'obj/gsromfs1_.c'. Stop. | 17:13.20 |
chrisl | Er, huh? I've just done a bunch of builds, and a clusterpush with the change..... | 17:14.28 |
fredross-perry | Iâve reset my mupdf repo to match golden, and added two commits. One relates to saving in iCloud, please take a look and see my previous email on that topic. Thanks. | 17:14.38 |
marcosw_ | chrisl: okay, must be on my end... | 17:15.12 |
chrisl | marcosw_: Let me double check | 17:15.22 |
Robin_Watts | fredross-perry: Just a mo, and I'll look. | 17:15.31 |
fredross-perry | thanks | 17:15.42 |
chrisl | marcosw_: no, it's definitely my mistake - I cannot fathom I didn't see the same error..... | 17:16.35 |
marcosw_ | chrisl: the cluster wonât find the problem I see, it does a âmake -j12 ; makeâ, so if the parallel make files the serial make will work. | 17:17.01 |
chrisl | marcosw_: I thought it was setup to record when parallel builds failed | 17:17.49 |
marcosw_ | think of it this way, you made the the parallel make fail 100% of the time instead of 2.5%. And now there is an error message, so all good. | 17:17.55 |
| chrisl: me too. wonder what happened. | 17:18.59 |
chrisl | marcosw_: revised commit pushed - should work | 17:19.23 |
marcosw_ | chrisl: will do. | 17:20.17 |
| chrisl: Iâll also look into the cluster parallel make failure lack of email issue. | 17:22.28 |
fredross-perry | bbiab | 17:27.57 |
marcosw_ | chrisl: what did the previous attempt to fix the build problem look like? I want to use it to test the cluster parallel make reporting. | 17:29.03 |
chrisl | marcosw_: did the previous attempt not break all builds for you? | 17:29.48 |
marcosw_ | chrisl: just parallel make. but I think I figured it out. | 17:30.05 |
| was it: âPS_ROMFS_DEPS=$(PSSRCDIR)$(D)psromfs.mak $(gconfig_h) $(iconfig_h) \â | 17:30.20 |
chrisl | Yes, exactly | 17:30.39 |
marcosw_ | now whatâs the cluster doing? | 17:33.11 |
Robin_Watts | fred: There is trailing whitespace in your last commit, line 280ish. | 17:34.51 |
| ignore me. | 17:35.09 |
| fred: OK, there are lots of whitespace problems with your last commit. | 17:36.09 |
| fred: Along with the wrong brace style in 2 places. | 17:36.31 |
| The actual code is iOS, so I'm not a great reviewer, but it looks plausible enough. | 17:36.59 |
jogux | Robin_Watts: I plan to look at the iOS changes at some point... it's just not very far up my rather long list :-( | 18:05.50 |
fredross-perry | is there a convenient way of searching the IRC logs, if youâve forgotten what day it was? | 18:09.51 |
Robin_Watts | fredross-perry: Top right of the logs is a google box. | 18:17.25 |
fredross-perry | Ah, I thought that was for the current page. Thanks. | 18:18.35 |
Robin_Watts | np. | 18:18.47 |
mvrhel_laptop | off for lunch. bbiaw | 19:25.53 |
| done for the day | 23:49.08 |
| Forward 1 day (to 2015/10/07)>>> | |