| <<<Back 1 day (to 2014/06/23) | 2014/06/24 |
kens | chrisl from the Very Sleepy profiles, using ppmraw at 72dpi, it does appear that the majority of the performance cost is in FAPI. In the old code zbuildfont1 (the main font routine I see called) comes in at 2.36 seconds. With FAPI zFAPIrebuildfont comes in at 12.77 seconds. I'ml;ooking for a difference of around 16 seconds so there's a huge chunk f that right there in the FAPI code :-( | 07:42.21 |
| I notice the customer hasn't replied further to my mail last night though | 07:43.12 |
chrisl | kens: well, as I say, if the larger portion of that time is spent in names_ref(), then the only solution I see is to pre-create a list of refs with the appropriate names at init time. | 07:54.03 |
kens | chrisl I'm not sure precisely where the slowdown occurs, isolating it to the FAPI code is about as far as I can go, I don't know hte code well enough any more. I'm not certain that much of the actual time is lost in names_ref though, about 2 seconds of the 12 seconds overall I think. I'm disinclined to go further with a file which the customer now says is *not* the originalbut has been 'cleansed' and 'scaled down'. | 07:56.12 |
| Given that I found places in the file where it switches font 14 times before settling on one, and the program ends up with > 5,000 objects left on the stack | 07:56.43 |
| I'm not convinced that there's any point in looking further at that file, to my mind the customer has damaged it. | 07:57.06 |
| I suspect that this file is simply a pathological case, given that for the other customer that reported a problem we now perform as well as 9.05 or better. It does seem like we generally suffered a hit with pdfwrite when we introduced FAPI, back in 9.00 but frankly its too late ot change now, and the number of bug fixes that would disappear if we tried to go back to the Artifex code makes me shudder | 08:00.03 |
| THat's pretty much the line I was drawing last night in my mail, and I plan to stick to that, unless they can provide a file whioch *hasn't* been meddled with and shows a performance penalty > 2 | 08:00.56 |
kens | coffees | 08:04.17 |
chrisl | kens: Hmm, unlike what I thought, FAPI is not needlessly rendering glyph bitmaps..... | 08:23.59 |
kens | :-( | 08:24.08 |
| TO be honest, I think that's what the 72 dpi ppmraw test suggested too | 08:24.27 |
| When the rendering/writing bitmap time is small I see that the total time is about 2.5 times slower than 8.71, and 25% slower than with -dDisableFAPI, which suggests a genuine performance hit for using FAPI. | 08:25.53 |
| On the other hand, we know there are a number of files which now work, and previously (ie 8.71 vintage) did not, so..... | 08:26.22 |
chrisl | Yes, I'm more bemused by pdfwrite seeming to suffer so badly | 08:27.00 |
kens | I think its because pdfwrite itself doesn't change much. For rendering devices the interpretation time isn't a big part of the total time, for pdfwrite, it really is, especially for simple and small files. | 08:27.49 |
| Like I said, if I keep the rendering time down (by using low resolution) I see about the same penalty with rendering as with pdfwrite. | 08:28.44 |
chrisl | Hmm, with ppmraw, 72dpi, we're very slightly slower with DisableFAPI running the PLRM PDF file......... | 08:33.46 |
kens | The customer file took 1.99 seconds with master and 72 dpi as opposed to 1.5'ish with 8.71 | 08:34.23 |
| That's pdfwrite timings | 08:34.38 |
| Wihtout DisableFAPI the time is 2.5 seconds | 08:35.26 |
chrisl | It does suggest that the time penalty is in font creation, rather than rendering | 08:35.34 |
kens | Yeah, I think that's the problem, and like I said, if you look in the file you will see it creates a font, creates another font, creates the first font again...... | 08:36.09 |
| Which is one reason I distrust it as a test file. Whatever the customer did to it, they broke it. | 08:36.30 |
chrisl | So, again, I think the two options would be to pre-create the name refs for the PS name objects we use in FAPI, or switch to lazy FAPI font creation | 08:37.35 |
kens | Well both sound kind of ugly. Do you think pre-creating the names would help much ? | 08:38.13 |
| In this file, lazy creation would probably help a lot (as they throw the fonts away so much), but I'm not at all convinced that owuld help the original, unmodified, file | 08:38.55 |
| WHich is one reason I'm reluctant to spend much more time on it, I really don't want to waste time on a breokn file, getting it to work really quick, only for the cutomer to turn round and say 'butt he original file is no faster' | 08:39.46 |
chrisl | No, I agree. Lazy creation would only help bonkers files - the ref pre-creation stands a chance of helping in general | 08:40.01 |
| I did look at speeding up the name hashing, but the gains I found were microscopic | 08:41.12 |
kens | My feeling is that we should drop this for now, and return to it only if/when the customer responds again. If they do I'llpress for the original file stressing the point that this file is broken and 'fixing' it will almost certainly not help their original program much | 08:41.22 |
| I suspect that they may decide the current code is 'good enough', but how canI tell ? I don't know what sort of performance teh real tesat file will exhibit | 08:42.17 |
chrisl | Sure, yes. I'm not jumping through hoops for this case. But I might hack together a proof of concept for the ref pre-creation and see if it does reap any benefits | 08:42.44 |
kens | Well, if you can do a quick and dirty hack for investigation that would be great | 08:43.23 |
chrisl | Assuming my memory is correct, and name refs are atomic objects, it shouldn't take long | 08:43.56 |
kens | I'm reasonably certain they are | 08:44.08 |
chrisl | So no need to worry about gc | 08:44.22 |
kens | Yeah I'm fairly certain that names aren't gc'ed | 08:44.34 |
| When you test it I would compare the low resolution rendering, to remove any pdfwrite strangeness | 08:45.04 |
chrisl | I'll have a poke at it when I get the cust 532 business out the way | 08:45.28 |
kens | Because its possible pdfwrite is still exercising the code in odd ways itself, especially when it comes to widths and stuff | 08:45.34 |
| OK thanks chrisl, I'm diving into colours now for a bit | 08:45.50 |
kens | may be gone some time | 08:45.58 |
chrisl | Good luck! | 08:46.02 |
Robin_Watts | jogux: So I can solve the 'jump to the end -> crash' bug. | 09:45.01 |
| but it's not ideal. In order to reproduce it, you have to put the caret in some text, then hit ctrl-end. | 09:45.30 |
jogux | ok. cool | 09:45.50 |
| it's a relatively niche piece of functionality anyway | 09:45.57 |
Robin_Watts | The problem is that the last thing in the file isn't text. | 09:46.00 |
| so you hit ctrl-end and basically nothing happens. | 09:46.18 |
jogux | ship it! | 09:46.25 |
| seriously. I think that's okay. if you were going to worry more about something, I'd worry that you up arrow often doesn't move you to the next bit of text above. | 09:47.01 |
Robin_Watts | Possibly, if ctrl-end is only going to navigate between text items, it should jump to the last 'text' item in the file, not the last selectable item. | 09:47.02 |
| jogux: OK. I will prepare a commit for what I have, and then look at the customer problem that came in yesterday. | 09:47.27 |
| I might open a bug for cursor movement not being correct. | 09:47.38 |
jogux | nods. to be honest I'm surprised we've not had any complaints about it. | 09:48.02 |
| albeit... the current release on iOS doesn't even support cursor keys, so people were complaining about that instead :-) | 09:48.30 |
| did we deliberately stall on the on the appstore releases of SO2, or are they stuck hoping someone tests them more? | 09:52.51 |
Robin_Watts | The latter, I suspect. | 09:53.25 |
Robin_Watts | runs while precommit test runs. | 10:05.55 |
| jogux: It's the 3rd commit on robin/master if you have time, thanks. | 10:06.11 |
jogux | I never noticed gitk's "color words" option for the diff view before. It's very good. | 10:41.36 |
| robin_watts: that commit LGTM, assuming ATS is happy. | 10:47.50 |
| robin_watts: that commit LGTM, assuming ATS is happy | 10:48.21 |
paulgardiner | jogux: re app store release - since finding and fixing that regression that causes a crash if you open in from another app, I need to upload a new version to testflight for testing | 10:56.55 |
jogux | ahh, right. | 10:57.07 |
paulgardiner | jogux: and on Play I'm discussing moving the old app between accounts with Miles | 10:57.53 |
jogux | cool. :-) does that need Miles to do much? | 10:58.31 |
paulgardiner | Moving to Skype | 10:59.12 |
| There is one other thing we should sort out before we release: clicking on Help still sends the user off to picsel.com | 11:22.21 |
| SO2 help goes here: http://www.picsel.com/docs/user-guide/smart-office/en-gb/2.3.0/iOS/tablet-hidpi/introduction | 11:26.28 |
| Maybe I should talk to Ron about getting that moved to an equivalient within the artifex domain | 11:27.14 |
jogux | paulgardiner: the smartoffice link on the artifex website takes you to picsel.com :-) | 11:28.22 |
paulgardiner | Hmmm | 11:29.10 |
| oh but to updated content | 11:30.23 |
| So perhaps we just need to update the help files too | 11:30.46 |
jogux | very possibly | 11:32.03 |
| at least removing the "Copyright © 2011-2013 Picsel International Ltd (in liquidation)" would be good :-) | 11:32.16 |
paulgardiner | :-) | 11:32.26 |
jogux | wonders how that website works. http://www.picsel.com/docs/user-guide/smart-office/ seems to display exactly the same content as Paul's url. | 11:33.30 |
| oh, except the links there don't work. odd. | 11:34.11 |
| http://www.picsel.com/docs/user-guide/smart-office/en-gb/2.3.0/iOS/tablet-hidpi/appendix#document suggests there's some autogeneration going on. | 11:34.53 |
paulgardiner | Some sneaky modrewrite maybe | 11:35.17 |
jogux | possible. Let's see if Ron can figure it out :-) | 11:35.46 |
Robin_Watts | crap. TGV test fails. | 11:35.52 |
jogux | :( | 11:36.04 |
rayjj | jogux: thanks for mentioning --color-words for git. I use vimdiff quite a bit and like that (it's there by default) | 12:03.57 |
| --color-words combined with -w help finding the "real" diff | 12:04.34 |
jogux | ahh. cool. I'd not quite made the link that it was also a command line option :-) | 12:04.58 |
kens | rayjj am I misunderstanding your proposed 'page neutral' change ? I'm thinking this will be available for inspection on all printer devices. | 12:05.23 |
rayjj | jogux: I use command line (msysgit) a LOT. Almost never use gui tools. | 12:05.39 |
kens | rayjj so once its in a release teh customer that prompted it won't need a 'special' build done for them | 12:05.49 |
rayjj | it's almost like having a real computer when I can use bash and all my unix tools | 12:06.28 |
| kens: I think that customer (534?) should step up to being able to do a build | 12:07.20 |
kens | is inclined to agree | 12:07.29 |
rayjj | kens: I'll finish putting the change in as we discussed (for printer class devices) | 12:08.02 |
kens | OK that's fine. I just wanted to make sure I was understanding correctly | 12:08.25 |
| I thought about writing an email this morning then thought maybe I had the wrong end of the stick | 12:08.45 |
paulgardiner | pedro_pc: I've just tried getting the GD suite plus SOG talking again. This time Good Share looked like it was allowing me to authentigate but then complained that it was delegated to GFE and that was missing. Great! | 12:08.59 |
rayjj | chrisl: BTW, thanks for all the work with Shinuhara-san | 12:09.13 |
pedro_pc | paulgardiner | 12:09.20 |
paulgardiner | pedro_pc: I could have one more go, this time starting with GFE, but we have just two keys left and I don't want to use them up if you need them | 12:09.41 |
pedro_pc | paulgardiner : it seems pretty broken just now | 12:09.49 |
rayjj | emails aren't annoying enough ??? | 12:10.03 |
chrisl | rayjj: no problem - I was surprised I got where I did in the time it took.... I *hope* it works for the customer! | 12:10.14 |
rayjj | chrisl: the gp_monitor_free change is part of it. Is that in HEAD ? | 12:10.57 |
paulgardiner | Shall I go for it and then contact Brad, or do you want to experiment with those last two keys? | 12:11.00 |
pedro_pc | paulgardiner: think we probably need to prod Good again. Want me to do it or do you? | 12:11.13 |
chrisl | rayjj: it's not there yet - it will be in the commit when I push it. I thought I'd hold off pushing it until the customer is happy | 12:11.37 |
pedro_pc | paulgardiner: I have suspicions about our login having expired (think you did too) | 12:11.55 |
paulgardiner | I'm happy to, but it's just whether to try this one more experiment (using up the keys) first. It sounds like the keys wont be much use to either of us so no loss. | 12:12.19 |
pedro_pc | I have apps on two devices already authenticated but Goodshare just isn't playing at all, either itself or thrugh GfE (it bombed out when I tried to attach a GoodShare doc with GfE mail) | 12:15.33 |
| so I don't specifically need the keys right now - if you want to experiment then feel free | 12:15.54 |
| we can get more issued if we need them | 12:16.04 |
rayjj | chrisl: OK. | 12:16.20 |
paulgardiner | pedro_pc: ok thanks | 12:18.08 |
pedro_pc | paulgardiner: I'm back using our GC / apps for now. I've been through 3 different reworks of the PicselViewer activity/app trying to resolve our single/multi-task issues with Good and I'm just about to try the last option (which is a bit of a hybrid). I've pretty much run out of other options, but I have a support query out with Good to see if they have any alternatives (or a roadmap to support singleTask activities) | 12:21.02 |
| I've managed to limp through most of the Good test sheet though, which seemed pretty much OK | 12:22.15 |
| (OK apart from the frequent crashes switching back-and-forth between activities due to the singleTask issue that is...) | 12:22.50 |
Robin_Watts | jogu, pedro_pc: OK, tgv tests pass. Updated commit on robin/master | 12:24.56 |
pedro_pc | takes a look | 12:25.17 |
jogux | robin_watts: sounds reasonable | 12:27.45 |
Robin_Watts | Ta. | 12:27.50 |
pedro_pc | yeah, LGTM | 12:32.15 |
paulgardiner | pedro_pc: Just sent my Brad email to you in case you want to add something. If not, let me know and I'll send as is. In general, if you have questions for GD, you can always send me an email which I'll forward on so that they see the artifex email address. | 12:41.40 |
pedro_pc | paulgardiner: cool, thanks. Just reading now | 12:42.25 |
| paulgardiner: that looks like a pretty decent summary of the state of play. GfE seems fine for me still (I can send/receive emails through their domain) but all interaction with GoodShare fails, as does any attempt to find servers. My GoodShare apps on android manage to authenticate ok but say "Failed to validate access' when I try to select the 'Datasources' folder (where all the shard services live...) | 12:46.54 |
paulgardiner | Presumably you've not uninstalled and reinstalled in the last two days, so you may well have seem the same problem with provisioning as I do. | 12:49.28 |
jogux | robin_watts: ok. pptx placeholder stuff. I believe I now understand what is wrong. | 12:53.22 |
paulgardiner | pedro_pc: Sent. I added a bit concerning your results | 12:54.05 |
pedro_pc | paulgardiner: ta. You're right about the auth too - it was last week I registered the last one | 12:57.34 |
kens | Marcos' | 13:00.39 |
| And I can't seemto silence it :-( | 13:00.50 |
pedro_pc | stealthbot | 13:01.18 |
jogux | kens: ignoring it has worked for me :-) | 13:02.43 |
kens | suspects Miranda is broken | 13:03.37 |
chrisl | My filter on it works, but it still makes the logs noisy | 13:03.43 |
paulgardiner | pedro_pc: am I remembering correctly?: we have been asked to perform two updates to new SDK versions, firstly to get to 1.6.x from 1.4.x and then another update because of an openssl problem. I can't find the email that informed us of the second SDK. | 13:06.53 |
pedro_pc | that's right - the email was forwarded by henrys I think | 13:07.22 |
| june 12th | 13:08.46 |
paulgardiner | ta | 13:10.41 |
pedro_pc | the UI also changed a bit (rebranding on their part) along with the SSL fix, so presumably just a cut of the current state of play from them | 13:10.58 |
paulgardiner | Did it require any chances to the Android app? | 13:11.40 |
| Other than using that SDK | 13:11.54 |
pedro_pc | no, just a drop-in replacement | 13:12.05 |
paulgardiner | Good good :-) | 13:12.21 |
Robin_Watts | at least in the conversion from raw -> html that should be easy. | 13:18.31 |
chrisl | Robin_Watts: I didn't see that - if so, he hasn't yet | 13:18.44 |
kens | Its in the logs today | 13:18.47 |
chrisl | Ah. I still think it would be better to send the messages to an unlogged channel - as we used to do with commit notifications | 13:19.32 |
Robin_Watts | chrisl: That would just mean I had to swap tabs all the time :) | 13:20.34 |
chrisl | Robin_Watts: do you really benefit from seeing all the cluster activity changes?? | 13:21.08 |
Robin_Watts | no. | 13:21.15 |
| Though I do like seeing notification of commits. | 13:22.08 |
| I guess I'd like to see notification of commits, and big red notifications when tests fail :) | 13:22.31 |
chrisl | I'd like it all to go a separate channel - except maybe the failure | 13:23.19 |
kens | would settle for being able to ignore it | 13:24.01 |
kens | thinks he found out where to mute it :-) | 13:53.23 |
henrys | rayjj: thanks. | 13:54.16 |
| FWIW I vote for chat only on IRC | 13:57.20 |
| chrisl, tkamppeter I meant to ask you guys about X11 and Mir, there must be some sort of compatibility mode with X windows right? | 13:59.12 |
| thatâs going to be a rocky change for ubuntu | 13:59.42 |
chrisl | henrys: I don't know much about Mir - I don't tend to pay much attention to that kind of unilateral nonsense...... | 14:00.34 |
kens | my mute still failing :-( | 14:03.18 |
rayjj | kens: can you have a look at my git repo (patch to add pageneutralcolor as a parameter to printer devices) ? http://git.ghostscript.com/?p=user/ray/ghostpdl.git;a=commitdiff;h=d89bb6d62a2178b8a4e428402418b963635864ff | 14:05.27 |
kens | Sure, give me a minute | 14:05.40 |
rayjj | kens: oops. that's NOT the right commit. I clusterpushed, but didn't commit first | 14:06.25 |
| kens: just a sec | 14:06.30 |
kens | OK | 14:06.33 |
rayjj | kens: OK. Here's the one I did the clusterpush on: http://git.ghostscript.com/?p=user/ray/ghostpdl.git;a=commitdiff;h=2fb610456b34ba57e257f954d41aa7adccafd9a6 | 14:07.38 |
kens | reading now | 14:07.50 |
| So the device parameter is available if we set -dGrayDetection on the command line ? | 14:10.00 |
| and not otherwise ? | 14:10.06 |
| Or rather it will always be false otherwise | 14:10.29 |
rayjj | kens: right. It returns false if GrayDetection was false (false means we think the page has color) | 14:11.28 |
kens | I guess that's reasonable, would there be a cost to just always doign GrayDetection ? | 14:11.46 |
rayjj | hmm... I suppose I should add documentation to the place where GrayDetection is mentioned | 14:12.08 |
kens | Documentation *somewhere* would be good, otherwise everythign looks just fine to me. | 14:12.31 |
rayjj | kens: yes, it costs to do the color monitoring. We turn it off as soon as color is detected | 14:12.43 |
kens | Right, then its a good idea to require GrayDetection | 14:12.56 |
rayjj | kens: mvrhel did this for customer 801 (that never became a customer) | 14:13.15 |
kens | It seems useful, we do get this request from time to time | 14:13.34 |
rayjj | kens: I'll do docs in a bit, then request a review to see if I'm adequately obtuse ;-) | 14:15.49 |
kens | :-) | 14:15.56 |
| No matter how clear you make it, someone will misunderstand.... | 14:16.14 |
jogux | robin_watts : let me know when you have brain time to give to the styling stuff. I'm fairly certain I know what's wrong, but I need to investigate more before finding the fix. | 14:19.08 |
Robin_Watts | jogux: I'm just writing an email to a customer now, but I can park that. | 14:19.37 |
jogux | let me start then | 14:20.21 |
| so there are two issues; one is what's wrong, one is why we didn't understand what was going on | 14:20.37 |
| or at least why I didn't, which hopefully has some overlap with why you didn't | 14:20.58 |
| in the edr dump, we have: | 14:21.11 |
| Group, idx = 77 type PH PlaceholderID='1' attr:(shared)0012 inline:(shared)0008 styles Shape id='1' manager=Drawing | 14:21.13 |
| hm. wait. no. I still don't understand this. darn. | 14:21.53 |
jogux | gives whoever designed this debug output a good kicking. | 14:22.08 |
Robin_Watts | jogux: Should I git blame? :) | 14:23.49 |
jogux | I think I understand why your debug output wasn't giving what we expected (or what I expected) - aiui, the output includes style rules that have already been partially matched, and where they have partially matched the parts that have matched are removed from the selector. | 14:24.12 |
Robin_Watts | Ah! | 14:24.30 |
jogux | and there's a completely separate function that deals with that part of the matching which doesn't have 'match' in it's function name. | 14:24.59 |
Robin_Watts | of course :) | 14:25.37 |
jogux | life would be so so much easier if management hadn't continually tried to get us to write a full web browser | 14:27.42 |
pedro_pc | is it broken specificity or just unfortunate processing order? | 14:27.53 |
jogux | pedro_pc: I'll get back to you on that. | 14:28.32 |
pedro_pc | :) | 14:28.36 |
jogux | the root cause is "hugely overcomplex code". | 14:28.45 |
pedro_pc | mm | 14:29.22 |
jogux | which sadly also has no debug and the comments are sparse and don't explain the high level principle of what it's doing :-S | 14:33.17 |
pedro_pc | mourns | 14:37.26 |
| I think a lot of that code has been 'optimised' over the last few years and lost some of the (more obvious) structure | 14:40.10 |
Robin_Watts | despairs at customer. | 14:41.59 |
| If you're going to send me a file that works with one version of the lib, but doesn't work with another version, TELL ME WHICH VERSIONS. | 14:43.08 |
paulgardiner | pedro_pc: the app list in our Good Control looks very different now. | 14:43.14 |
| com.picsel.smartofficegd is missing | 14:43.35 |
| pedro_pc: what should I be using now if I want to provision from our own GC? | 14:44.51 |
pedro_pc | paulgardiner: that was the change I was doing yesterday - trying to rationalise things and make it more apparent what were artifex dev apps and what were legacy picsel ones. The 'Artifex Dev SOG' is the equivalent. I can recreate a com.picsel.smartofficegd but I was finding it confusing between the different iOS and android identifiers on the portal and in the GC so wanted to clean up the naming to make things a bit more obvious | 14:46.32 |
| paulgardiner: for the dev SOG I changed the Good appID to 'com.artifex.dev.sog' | 14:48.16 |
| and iOS bundle ID to com.picsel.smartofficegd | 14:48.49 |
paulgardiner | What are the PSO2 ones? | 14:49.16 |
| Items 4,5 and 6 is in the list? | 14:49.38 |
pedro_pc | paulgardiner: removed (as it was only a build peculiarity that resulted in com.picsel.smartoffice2 apps for good) | 14:50.12 |
| do you want me to add that back in as the bundleID for the dev builds? | 14:50.40 |
paulgardiner | Not sure. I'm still in the process of understanding | 14:51.20 |
| PSO2 ones are still there | 14:51.44 |
pedro_pc | basically we only have one android and one iOS Good app, but we had a variety of identifiers both in our GC and in the Partner portal, soe overlapping. | 14:52.08 |
| I wanted to rationalise that to just having what we need , now that we understand more | 14:52.45 |
paulgardiner | iphone_postlink.py has "com.picsel" hardwired. I could change that, but it might be better to stick with it as the prefix, unless there are other issues | 14:53.00 |
| pedro_pc: yeah I'm with you on that. Good to rationalise. I'm just not understanding what's left. | 14:53.44 |
pedro_pc | so there is now only one SOG app in our GC (for development) and it is a different appID from the marketplace build | 14:53.47 |
| paulgardiner: do you mean hardqired for the Good ID or the iOS package ID? | 14:54.12 |
paulgardiner | both | 14:54.20 |
pedro_pc | that's a bit of a pain, but I'll switch it back then. I just wanted to have a clear distinction between dev and marketplace apps so that we avoid issues with the NOC | 14:55.16 |
paulgardiner | Maybe the site is showing us different versions of that list. I can still see "Picsel Smart Office 2 (Android) | 14:55.17 |
pedro_pc | They are the old Picsel entries | 14:56.00 |
paulgardiner | Looks like postfixing with "dev" has been used in the past | 14:56.02 |
jogux | paulgardiner: I added postfixing with dev? or is it in other places too? | 14:57.50 |
paulgardiner | Just noticed iphone_postlink.py defaults to com.picsel.smartoffice2gd, and changes that to com.picsel.smaroffice2gddev if an option is added to the build | 14:58.35 |
jogux | ah, that wasn't me :) | 14:58.44 |
pedro_pc | paulgardiner: those apps have only legacy bindings though- not the current FileTransfer | 15:01.07 |
paulgardiner | Which ones? | 15:01.29 |
pedro_pc | the PicselSmartOfficeDev ones | 15:01.43 |
| Picsel Smart Office 2 (Development) only has the Picsel Document service | 15:02.28 |
| I'd be very much in favour of dumping all those existing app identifiers (other than the marketplace one) and fixing on one for development | 15:03.08 |
paulgardiner | Oh. Are the PSO2 apps ones that the builds system can still create? | 15:04.42 |
pedro_pc | they're all just variants of the Good SO2 build with different policies/service bindings | 15:06.06 |
| none of which are actually much use to us | 15:06.17 |
paulgardiner | Right. so we could dump those and reuse their GD App IDs for something else | 15:07.05 |
| ... in particular it allows prefix of "dev" for our development version | 15:07.27 |
| Or am I still confused? | 15:07.33 |
pedro_pc | or dump them - the Good IDs aren't registered; they're just the legacy IDs Picsel were using on their own GC | 15:07.45 |
paulgardiner | Could do, but I was thinking +"dev" was preferable to s/picsel/artifex/ for the dev versions | 15:08.38 |
pedro_pc | at a minimum I think we should have the same Good app id on both iOS and android, unless we plan on supporting different services | 15:08.58 |
| ok, lets dump the artifex bit and stick with com.picsel.smartoffice2, smartoffice2dev? | 15:09.33 |
| (sorry, smartoffice2gd, smartoffice2gddev for the package IDs) | 15:10.17 |
paulgardiner | Yeah. That looks good to me | 15:10.34 |
| What's "GoodShare service test app (android)"? | 15:11.26 |
pedro_pc | picsel's local GoodShare trial - they set up a share server locally | 15:11.53 |
| I'm presuming that was running the 'Picsel Document Service' | 15:12.13 |
Robin_Watts | So, the customer appears to be complaining that the MIPS build I do for them gives blurry results, and the non-mips build works OK. | 15:13.54 |
| but the MIPS build is an fvdk build, and the non-mips build is a page-view one. | 15:14.17 |
paulgardiner | henrys: No general meeting today, I know, but are we discussing SO? | 15:14.25 |
jogux | robin_watts: do we have a firm definition for 'blurry'? | 15:14.47 |
Robin_Watts | They sent me a blurry photo :) | 15:15.04 |
henrys | paulgardiner: sure why donât we go ahead and have an SO meeting. I heard from michael and heâs made a bit of progress and was planning to contact jogux | 15:15.32 |
jogux | did that help? :) | 15:15.33 |
Robin_Watts | of 2 projections. Supposedly the one on the left is "OK" and the one on the right is "not OK", but I can't tell :) | 15:15.39 |
jogux | robin_watts : go, I'll bite, let's see the photo :) | 15:16.07 |
henrys | paulgardiner: how is the good stuff going? | 15:16.16 |
jogux | henrys : that's good news. apart from the contacting me bit :) | 15:16.19 |
Robin_Watts | jogux: essentially, no. They haven't given me enough information to reproduce it. Let me mail the photo. | 15:16.25 |
pedro_pc | paulgardiner: So... what I'm planning to do is remove all of the Picsel Smartoffice 2 entries, the Good Share test app, the 'Smartoffice for Good' entry (the com.artifex one I created) and update Artifex Dev SOG to use com.picsel.smartoffice2gddev as the Good app ID and com.picsel.smartoffice2gddev for the ios bundleID and com.picsel.tgv.app.smartofficegd2dev as the android packagename | 15:17.01 |
Robin_Watts | Is the fb device something I can enable on an ubuntu box ? | 15:17.09 |
paulgardiner | pedro_pc: sounds good to me | 15:17.30 |
pedro_pc | Robin_Watts: you want an iPAQ for that | 15:17.32 |
paulgardiner | henrys: iOS is looking very close now | 15:17.43 |
| ... although I suppose it was last week, so.... | 15:18.01 |
henrys | paulgardiner: you mean for the upgrade - sans heartbleed ;-) | 15:18.15 |
jogux | robin_watts: not great. deliberately without reading the email. I'd say the one on the right is showing signs of pixelation. | 15:18.39 |
paulgardiner | henrys: while testing I found a crash that affected both SOG and SOT. That's fixed now. I also got Picsel's GD service test program working and used that to test iOS SOG and it all seems to work. | 15:19.11 |
henrys | jogux: a quote from michael in case you want to warn him hell awaits : Part of the issue is that the pie-chart code seems to assume that the colors of the chart will follow a particular theme palette in order for each segment and does not allow for out of order color selections with the palette | 15:19.15 |
pedro_pc | android too - I've been through most of the Good tests required for Veracode and they have been pretty positive. Still trying to resolve our issue with not being able to run as a single task on android | 15:19.22 |
jogux | henrys : oo, joy. | 15:20.04 |
Robin_Watts | jogux: The page content here is a full page image. | 15:20.20 |
jogux | robin_watts : my immediate thoughts would be 'image subsampling might be kicking in, either due to explicit configuration, or lack of RAM' | 15:20.34 |
paulgardiner | henrys: one of the tests GD suggest is to check communication with Good Share. We haven't managed to do that, but it looks as though it's not the app, but rather the set up that Brad and Richard have given us | 15:20.43 |
Robin_Watts | (I think). Their test file consists of "real page", "image of what that real page should look like". | 15:20.49 |
| Yes, subsampling had occurred to me too. | 15:20.56 |
paulgardiner | One last step I looked into today was performing the second not-because-of-heartbleed-honestly SDK update. That's run me into some new hell but I think I'm through it. | 15:21.53 |
henrys | paulgardiner, pedro_pc are all outstanding questions sent to brad and richard? | 15:21.57 |
paulgardiner | henrys: yes, I believe so. | 15:22.14 |
Robin_Watts | hmm. building with their flag means it builds a picselapp.a, building without it means it builds a fileviewer exe. | 15:22.19 |
paulgardiner | henrys: there was something else I wanted to bring up | 15:22.27 |
henrys | uh oh | 15:22.33 |
Robin_Watts | I think I need to nobble that flag a big so I get the same thing built on x86. | 15:22.35 |
paulgardiner | there's very little stopping us doing releases of SOT on Play and the app store. | 15:22.53 |
pedro_pc | henrys: paul emailed Brad today about GoodShare. I have a gdsupport query in progress abou their lack of support for android singleTask activities | 15:22.56 |
jogux | robin_watts : there's a small chance building with -define=BETA_RELEASE will log useful stuff. also building with TGV_API_LOGLEVEL=1 or something might also log useful info about their environment. | 15:23.36 |
henrys | paulgardiner, pedro_pc I havenât been following how long has the ball been in there court? | 15:23.37 |
paulgardiner | but one thing still nto addressed is that the SOT help button opens a browser on picsel.com and shows pages with "Picsel" in the logo | 15:23.40 |
| henrys: also the Artifex web site's SOT link goes to picsel.com | 15:24.05 |
henrys | paulgardiner: oh let me look | 15:24.37 |
paulgardiner | The artifex web site link at least goes to pages that have been updated with s/Picsel/Artifex/ | 15:25.14 |
| but not the SOT help files. | 15:25.20 |
henrys | what the hell. Looks like we need a little QA in San Rafael ;-) | 15:25.21 |
paulgardiner | :-) | 15:25.32 |
henrys | although it does say âArtifexâ smart office when you get to picsel.com | 15:25.56 |
paulgardiner | henrys: yeah there is does, but not on the SOT help files | 15:26.50 |
henrys | but why use the picsel host name | 15:26.53 |
| ? | 15:26.54 |
paulgardiner | exactly. | 15:27.01 |
| Is this something to point Ron at? | 15:27.10 |
henrys | Iâll talk to miles about that I guess ron can straight it out | 15:27.13 |
| right | 15:27.14 |
paulgardiner | If we're sure we intend and are able to move it ontohttp://www.artifex.com then we can update the app to assume that before I create the release test version | 15:28.29 |
pedro_pc | paulgardiner: I presume you want to keep the iOS test services app on the GC? | 15:28.34 |
paulgardiner | pedro_pc: yes please | 15:28.44 |
| pedro_pc: could keep the AppKinetics one too if that's someting you've been using for testing | 15:29.26 |
pedro_pc | nods - I'm using that most of the time. | 15:29.48 |
jogux | henrys : fwiw I'm currently looking at a powerpoint display issue (the one Robin was previously looking at). It's proving tricky. ATS work is paused for just now, when I get back to it next big step is getting the windows build clients going still. | 15:30.07 |
pedro_pc | Good Pro looks interesting too - haven't seen them advertising it but noticed it on android Play store. Looks like the GfE replacement | 15:30.20 |
paulgardiner | henrys: just brought that up because I'll have to hold off creating a new testflightapp version until we are sure what the new url will be | 15:30.21 |
henrys | paulgardiner: Iâll send him email right now. does picsel.com resolve to the hosting Robin_Watts arranged? | 15:32.03 |
Robin_Watts | henrys: I didn't arrange any hosting. | 15:32.41 |
| for picsel.com | 15:32.51 |
| picsel.com goes to fasthosts.com, I think. | 15:33.02 |
| Miah sorted that out to get the original picsel stuff back up and running. All the log in details were passed to miles/ron for that (and they took over paying for it last week too) | 15:33.32 |
chrisl | It was planned to move it to Godaddy but I don't know if that went anywhere yet | 15:33.39 |
henrys | anyway they can sort it. Ron and Miles | 15:35.07 |
jogux | paulgardiner: I suspect we shouldn't use it as a reason to hold up the release - picsel.com can always be changed to redirect to the new location (which we'll need to do anyway probably for people using older builds) | 15:36.29 |
henrys | paulgardiner: you canât change testflightappâs host name? | 15:36.31 |
paulgardiner | henrys: this is where the SOT help button takes you: http://www.picsel.com/docs/user-guide/smart-office/en-gb/2.3.0/iOS/tablet-hidpi/introduction | 15:36.32 |
henrys | for each test? | 15:36.35 |
| paulgardiner: ugh that stll has the logo. | 15:37.33 |
paulgardiner | henrys: yes exactly | 15:37.43 |
| jogux: true, although if we can decide now that we will soon ensure that the same url with picsel replaced by artifex will be made to work soon, I could make the change before generating the test version | 15:39.00 |
Robin_Watts | If we can change the Copyright statement at the bottom of every page, and the "picsel" at the top to "Artifex" we're fine, right? | 15:39.00 |
| I have the login details for the fasthosts stuff if you want to do this... | 15:39.28 |
henrys | Iâm sure this is all content managed and not just static pages so it may not be so easy to change. | 15:40.09 |
paulgardiner | Me? Yeah could do. | 15:40.14 |
| So generally we are thinking don't worry about the url being picsel.com, just get the content right? | 15:40.39 |
henrys | paulgardiner: it will change soon but I donât think we should hold stuff up for the hostname change. | 15:41.42 |
paulgardiner | Okay. | 15:42.34 |
henrys | paulgardiner: let not have both you and Ron doing it. Should I have him do it or do you want to change it? | 15:42.40 |
paulgardiner | henrys: if he's happy to do it that suits me | 15:42.57 |
henrys | paulgardiner: okay | 15:43.05 |
| paulgardiner: I wouldnât even hold up work for the content. | 15:43.46 |
paulgardiner | henrys: sure the content can change under the url. It was just the url change I was considerin holding for | 15:44.29 |
henrys | paulgardiner: well weâll just leave it and the new name will just be an alias to the same ip address in the future right? So your link will still work. | 15:46.39 |
paulgardiner | yeah or a redirect so that the user doesn't even see the initial url | 15:47.29 |
henrys | Robin_Watts: looks like youâve been bogged down in OEM support? | 15:48.38 |
Robin_Watts | henrys: No, I was looking at pptx stuff to see why styles didn't match. | 15:49.19 |
| I found something we were doing wrong, but my attempt to fix it just got me very confused. So I passed that to Joseph, and he is now handling the confusion side of things for me :) | 15:49.53 |
| Then I looked at Smart Art (a missing feature), and it's very clear that that's a huge chunk of work that can't possibly be done by september, so I moved on. | 15:50.38 |
henrys | Robin_Watts: okay Iâve just seen a customer emails. | 15:50.47 |
| s/a/a few | 15:50.55 |
Robin_Watts | I fixed a crash this morning, and then I got bogged down in customer emails :) | 15:50.57 |
| I missed the customer emails - they ended up in my spam filter for some reason. | 15:51.12 |
| Marcos prodded me yesterday about them, which is when I found them. | 15:51.25 |
| so that's what I'm doing now. | 15:51.43 |
henrys | jogux: it will be great when ATS is running and we can hand that off to marcosw or am I being optimisic it wonât need a lot more fiddling? Presumably that would free you to work on SOT bugs. | 15:52.16 |
Robin_Watts | henrys: ATS is running enough for day to day use now. | 15:52.48 |
jogux | henrys : there's more we want working that's easier to do when it's here, but I've paused look at it to work on SOT bugs | 15:52.55 |
Robin_Watts | but there are further aspects of it that would help us. | 15:53.26 |
| I'm afraid it was me that pulled Jogu away from ATS. | 15:53.53 |
jogux | the state it is currently in will not work well in marcos's setup unfortunately, as we'll lose time whenever his box gets rebooted | 15:54.19 |
| and that seemed to happen frequently enough to be annoying | 15:54.38 |
Robin_Watts | It became clear that I just wasn't making meaningful progress on the styling problems, and it seemed likely that he'd resolve it faster than me. | 15:55.01 |
jogux | plus his upstream bandwidth isn't great :( | 15:55.09 |
Robin_Watts | The styling problems have potential to be a really good win for us. | 15:55.22 |
jogux | robin_watts : definitely. | 15:55.39 |
henrys | jogux: is there more bandwidth at emobix? I could probably convince Miles we need a few more hours to move along both the styling stuff and ATS, but I donât know if more people will help. | 15:56.25 |
jogux | henrys : ATS is actually currently at my house - the upstream is ~13Mbps. | 15:56.45 |
paulgardiner | Ah! I've forgotten my artifex gmail account password. Wanted to check the spam trap for customer emails | 15:56.57 |
henrys | jogux: are saying marcosw location needs to be reconsidered for the long term solution? | 15:57.37 |
Robin_Watts | jogux: I think henrys meant developer bandwidth :) | 15:57.39 |
jogux | ah, oops | 15:57.50 |
henrys | jogux: like Robin_Watts said, sorry about that bad choice of words given the context | 15:58.06 |
paulgardiner | Who's the admin for our gmail accounts? | 15:58.11 |
henrys | paulgardiner: Iâm one of them | 15:58.24 |
Robin_Watts | paulgardiner: marcosw I think. | 15:58.25 |
jogux | ah, oops. :-) emobix is probably fairly fully loaded just now. | 15:58.25 |
| henrys : long term I'm not sure. I suppose we need to try it. | 15:58.51 |
| henrys : is there a reason you're keen to make progress on ATS? | 15:59.14 |
| actually that said I am vaguely keen to move it as marcosw's machines have more power. hmm. although no one is complaining about the speed too much :) | 16:00.08 |
Robin_Watts | The cluster machines at marcosw's house shouldn't reboot that often (except when he turns them off to save power,and we can tell him not to do that for the ATS one :) ) | 16:00.10 |
paulgardiner | henrys: can you kick off a password reset for paul.gardiner@artifex.com? | 16:00.11 |
henrys | paulgardiner: sure | 16:00.28 |
paulgardiner | ta | 16:00.31 |
pedro_pc | has to pop out - back in 20 mins | 16:00.44 |
henrys | jogux: it does seem like something we (Artifex) should get under control. | 16:01.46 |
jogux | henrys : ah, fair enough. hm. | 16:02.09 |
henrys | and gets marcosw more involved, hopefully freeing you. But that may be wishful thinking ;-) | 16:03.22 |
Robin_Watts | Traditionally ATS admin has seemed 'hairy' to me. | 16:03.57 |
jogux | ATS isn't currently taking up any noticeable amount of my time | 16:04.03 |
Robin_Watts | It's all screen sessions, and 'start this then this, then this...' etc. | 16:04.14 |
| The more jogux can simplify that before he hands it back to us the better. | 16:04.30 |
jogux | I'd offer to image the current VMs and send them to Marcos, but I'm not sure that actually really solves anything. | 16:05.33 |
henrys | jogux: I guess it doesnât make that much difference. I do have to worry about Artifex functioning independently of Emobix and ATS seems to be a large hole, I realize there are others. | 16:06.52 |
jogux | I am personally a rather big point of failure just now. You're right that I need to fix that. | 16:07.40 |
| well, sort of. worst case if something happens to me is you have to bribe mace to go get the machine from my house and set it up somewhere else :) | 16:08.15 |
| oh, Matt knows ATS too actually | 16:08.48 |
| (mattchz) | 16:08.55 |
| henrys : anyway. I get your point. Let me get this styling stuff at least to a point where I could hand it back to Robin then try and focus on ATS. | 16:09.35 |
henrys | jogux: sounds good | 16:11.33 |
jogux | manages to alter Robin's doc to break things in a new way (which is good). | 16:26.29 |
pedro_pc | it is? ;) | 16:26.59 |
jogux | new way is more obvious and easier to debug :-) | 16:27.12 |
| robin_watts : ooo, weird. styleinternal API's intitialContext doesn't seem to be getting called for the group with type == PH. That explains some things. | 16:33.22 |
| ahha, because manager=Drawing means it's skipping 'normal' layout for these groups. | 16:35.27 |
pedro_pc | had vaguely expected the 'Manager' processing to handle setting up default contexts for things like slide masters | 16:43.22 |
jogux | pedro_pc : I'd expected the usual style inheritance rules to apply :-S | 16:44.45 |
| just for good measure, we'd now inside the bounds of the AUTOSHAPES_IN_LAYOUT ifdef | 16:45.19 |
Robin_Watts | special case code is always a pain in the ass :) | 16:52.12 |
| So, the subsampling threshold is set to 1000 in page-view | 17:00.26 |
| and the page image is 1280x800 | 17:00.37 |
jogux | so in page view it'd get subsampled to 640 x 400 ? | 17:01.04 |
Robin_Watts | yeah. | 17:02.41 |
| I can't see immediately why it would be better in one of their library versions than the other though... | 17:02.59 |
jogux | I'm presuming they're not running both binaries on the same device? | 17:04.52 |
Robin_Watts | one is mips, and one is ARM, so no :) | 17:05.05 |
jogux | :-) | 17:05.14 |
| do we subsample if the memory allocation fails for the original size? | 17:05.35 |
Robin_Watts | I don't *think* so. | 17:06.05 |
jogux | albeit 640x400@16bpp is 600K, so it's not exactly an outragous amount of memory (you'd hope) | 17:06.10 |
| robin_watts : do you want to know about Pal_Properties_dynamicSubsampling ? :-) | 17:07.59 |
Robin_Watts | jogux: aarg. flashbacks! | 17:08.17 |
jogux | robin_watts: and PicselConfigFV_DynamicSubsampling | 17:09.22 |
| I'm not really sure I'm helping | 17:09.32 |
| afaict that's not set differently in pageview vs fvdk, unless their alien is setting it in one. | 17:10.46 |
Robin_Watts | The fvdk/pageview difference may have been a mistake of mine. | 17:11.19 |
| I now think both are pageview. | 17:11.37 |
jogux | oh, but one is off that ancient release branch? | 17:11.53 |
Robin_Watts | both are off the current release branch, I think. | 17:12.09 |
jogux | oh. eep. so this is arm vs mips? | 17:12.18 |
Robin_Watts | yeah. | 17:12.30 |
| potentially at least. | 17:12.46 |
| I haven't ruled out customer error here :) | 17:12.54 |
| or robin error. | 17:12.59 |
jogux | :) | 17:13.11 |
| it may be worth giving them BETA_RELEASE TGV_API_LOG_LEVEL=1 builds of both, and running them through. iirc the output includes the build info, might output useful debug, and may also reveal differences in the alien between the two devices. | 17:14.03 |
Robin_Watts | jogux: At the moment they haven't actually confirmed exactly what two libraries I'd have to give them builds of :( | 17:15.27 |
jogux | :( | 17:15.38 |
| robin_watts, pedro_pc : I'm torn between making the drawing code process style contexts as 'normal', and going back and looking at a different approach. Maybe I'll just try the former and see what it breaks/fixes. | 17:23.41 |
Robin_Watts | Do we have any idea why it's currently handled in a special way? | 17:24.15 |
jogux | it's an autoshape | 17:24.30 |
pedro_pc | jogux: seems worth getting normal processing working, but I suspect it'll be a constant stream of minor niggles and issues for a while | 17:25.52 |
jogux | pedro_pc : I really meant keep all the special processing, just add the normal processing of stylecontexts into it. | 17:26.19 |
Robin_Watts | I mean, what do we supposedly gain by having used the 'special' processing? | 17:26.24 |
jogux | robin_Watts : a huge amount of code with comments that say 'MS do <x>' | 17:26.45 |
Robin_Watts | ah, right, that makes more sense, I think. | 17:26.48 |
pedro_pc | I'm sure with some reasonable thought the non-edr objects could have used the same style model but perhaps different low-level iterators | 17:26.50 |
jogux | pedro_pc : it's normal edr objects (ish), just processed in a special way. | 17:27.21 |
| I think basically what we're seeing is autoshapes use to be processed in the agent, but that means they weren't edittable, so that code was essentialy moved out of the agent and put into layout | 17:27.55 |
Robin_Watts | So "dynamic subsampling" basically means "don't do static subsampling", as far as I can see. | 17:29.21 |
Robin_Watts | bets mattchz is hiding under the desk. | 17:30.21 |
pedro_pc | jogux> yeah, the boundary is a bit hazy for me - edr-defined objects vs compact path/table/text etc | 17:30.44 |
jogux | robin_watts : he's off today ;-) | 17:30.59 |
Robin_Watts | ah, well timed :) | 17:31.13 |
jogux | if you stick uprintfchar() in appropriate places in the subsample code, they should output in BETA_RELEASE builds too | 17:34.19 |
pedro_pc | paulgardiner: looks like our trygoodnow subscription was just renewed... | 17:40.10 |
jogux | hometime. will experiement with applying style contexts as normal tomorrow. | 17:42.33 |
paulgardiner | pedro_pc: ah good. Pity I used all the keys! :-) | 17:47.56 |
pedro_pc | they've given us half a dozen new access keys | 17:49.24 |
| so existing apps seem to have expired | 17:49.41 |
pedro_pc | foods | 17:56.17 |
henrys | marcosw: do you have the duplex option on the control panel set to on? | 18:06.46 |
marcosw_ | henrys: I'm not at home but I don't think so. | 18:40.20 |
henrys | marcosw_: duplexing pcl is quite a puzzle | 18:41.19 |
marcosw_ | having worked on the duplex feature in the pcl device I agree | 18:42.10 |
henrys | marcosw_: pdf got it right languages should know nothing about duplexing | 18:42.58 |
mvrhel_laptop | henrys: sorry I missed the SOT meeting this morning | 19:40.13 |
| finally made it here. of course the bags did not make it :( | 19:40.26 |
henrys | mvrhel_laptop: ugh | 19:40.32 |
mvrhel_laptop | I hate united airlines | 19:40.37 |
henrys | mvrhel_laptop: I thought it was me, they seem to be always late | 19:41.23 |
mvrhel_laptop | anyway I made a bit more progress on some of this chart stuff. I have been spending a bit of time studying this chart content and how it is parsed. So the use of the palette and the ignoring of the custom colors actually may be due to a parsing issue in an earlier part of the code. Digging further today. | 19:44.02 |
| need to get my visual studio 10 updated with rock-n-scroll | 19:44.21 |
| driving me crazy not having it | 19:44.22 |
| Robin_Watts: what was it that you installed in yours? I recall you found something better | 19:44.37 |
Robin_Watts | mvrhel_laptop: Urm. let me look | 19:56.34 |
| Productivity power tools | 19:57.13 |
mvrhel_laptop | Yes that was it | 19:57.20 |
| Thanks Robin_Watts | 19:57.23 |
| ok that is much better. | 20:04.32 |
Robin_Watts | mvrhel_laptop: So, how goes SOT ? | 20:04.35 |
mvrhel_laptop | Robin_Watts: so far not too bad thanks to your hard work getting the test-shell set up in visual studio | 20:04.59 |
| I think I have the possibility of fixing a bug or two in the chart area | 20:05.13 |
Robin_Watts | cool. | 20:05.22 |
mvrhel_laptop | The open xml stuff format makes things a bit easier to see where things are going wrong too | 20:05.53 |
| I can see where the chart colors are getting missed during the parsing | 20:06.13 |
| and the chart stuff is pretty self contained | 20:06.27 |
| in the code | 20:06.30 |
Robin_Watts | you're clearly having a better time of it than me :) | 20:06.44 |
mvrhel_laptop | which makes it not too bad. but it is slow going just because it is new | 20:06.46 |
| well I think that is just because this is a good first problem to tackle | 20:07.22 |
| Robin_Watts: anything to do with the code at a higher level or building would be a big problem for me | 20:07.56 |
| oh this open xml sdk productivity tool is nice. I have been unzipping and looking at the xml in an editor | 20:09.21 |
| this is nice | 20:17.09 |
| http://visualstudiogallery.msdn.microsoft.com/450a00e3-5a7d-4776-be2c-8aa8cec2a75b | 20:17.11 |
Robin_Watts | bloody hell, yes! | 20:19.34 |
mvrhel_laptop | although I hate that the xml content is all one big line | 20:24.15 |
| I mean, come on, it could have line feeds and tabs | 20:25.03 |
| I think with some fixes on my editor that will be fixed | 20:33.16 |
| actually you have to go to edit->advanced->format selection | 20:49.24 |
| then everything looks reasonable | 20:49.36 |
henrys | darn it: http://arstechnica.com/tech-policy/2014/06/faa-grounds-amazons-drone-delivery-plans/ | 21:39.12 |
| Forward 1 day (to 2014/06/25)>>> | |