| | <<<Back 1 day (to 2014/03/23) | 2014/03/24 |
| tor8 | Robin_Watts: morning. | 12:07.03 |
| Robin_Watts | Morning | 12:08.57 |
| | Still in git hell here. | 12:09.02 |
| Jogux | :( | 12:09.08 |
| tor8 | oh :( | 12:09.15 |
| Robin_Watts | or at least, still trying to figure out what we should be aiming for. | 12:09.19 |
| tor8 | filtering taking forever? | 12:09.21 |
| Robin_Watts | tor8: Still trying to figure out the filter we should be using. | 12:09.32 |
| tor8 | I think someone mentioned doing a sed on the fast-export script? | 12:09.56 |
| Robin_Watts | tor8: yeah, not easy though, as the fast-export script doesn't contain any "delete the resources" things we can edit out. | 12:10.33 |
| henrys | quiet for a Monday | 12:10.53 |
| Robin_Watts | The fast-export script seems to be of the form: THIS IS COMMIT NUMBER x, THIS IS THE COMMIT MESSAGE..., [Optionally (PARENT IS X)*], LIST OF CHANGES | 12:12.01 |
| | IF no parent is given, it's assumed that the parent is x-1. | 12:12.17 |
| tor8 | Robin_Watts: can you restate what it is you need to do? | 12:12.31 |
| Robin_Watts | If 1 parent is given, then that's a branch, if 2 are given, it's a merge. | 12:12.36 |
| | tor8: Well, that's what we're trying to be sure of at the moment. | 12:12.48 |
| | We had *hoped* that at every tag within in epage, the same tag would have been applied in resources. | 12:13.09 |
| tor8 | and we have a new git with both the epage and resources modules in one? | 12:13.16 |
| | or two separate gits, with common tag names? | 12:13.22 |
| Robin_Watts | So to rebuild a given tag, we could have checked out a given tag and rebuilt, and we'd have been fine. | 12:13.38 |
| | We've built a new git with both epage and resources moduls in it. | 12:13.52 |
| henrys | good morning norbertj | 12:13.54 |
| Robin_Watts | BUT... the tags don't extend across both modules. | 12:14.02 |
| tor8 | cvs tags are per file, so any cvs-to-git conversion is going to have to make guesses | 12:14.17 |
| Robin_Watts | Yes. | 12:14.24 |
| norbertj | Good morning henrys, here it is already 12:14 | 12:14.28 |
| Robin_Watts | And we've ended up with some tags not having any resources in at all. | 12:14.45 |
| norbertj | I think I found a fix for the jpeg problem (just attached a patch for it to 695100). | 12:14.48 |
| henrys | norbertj: yes I thought that right after I typed it. 6:14 am here | 12:14.49 |
| norbertj | henrys: what do you think? and Hin-Tak? | 12:15.13 |
| tor8 | Robin_Watts: is this git somewhere on casper (so I can poke around and look and maybe think of something)? | 12:15.33 |
| Robin_Watts | tor8: It's on peeved. | 12:15.44 |
| | In /home/robin/PicselCVS/ | 12:15.58 |
| norbertj | henrys: 6:14 ? early riser ;) | 12:16.07 |
| tor8 | sot.git? | 12:16.25 |
| Robin_Watts | Your best bet is to copy sot3-2.git.tgz | 12:16.28 |
| | and unpack it yourself. | 12:16.34 |
| | cos I'm about to run another filter on sot.git to remove "Description:" lines. | 12:16.53 |
| tor8 | is peeved behind a modem or something? I'm getting < 100KB/s transfer speeds :( | 12:20.21 |
| Robin_Watts | peeved is in rays office. | 12:21.11 |
| tor8 | henrys: Robin_Watts: not sure who manages the DNS, but could we point mujs.com to a webpage hosted on casper? | 12:21.15 |
| | Robin_Watts: ah, that might explain it :) | 12:21.21 |
| Robin_Watts | tor8: ray normally handles DNS I think, but yes, we can certainly put mujs.com on casper. | 12:21.55 |
| | (belated) Morning henrys. Quiet here, but the phone has been burning between me and paul :) | 12:22.24 |
| henrys | Robin_Watts: did miles put your number on the app store? ;-) | 12:23.01 |
| Robin_Watts | henrys: no, we're trying to figure out this epage git stuff. | 12:23.20 |
| | One possibility is that for every tag, we should run back along the history until we find resources, and we should use that version there. | 12:24.45 |
| | Another possibility is that in resources files are never (or rarely) ever changed. | 12:25.25 |
| | in which case we could construct the 'latest' resources and use that everywhere. | 12:25.38 |
| | paulgardiner is looking into that later possibility now. | 12:25.51 |
| tor8 | Robin_Watts: if you start with separate epage and resources gits, for each tag in epage find the right set of resources (by searching back from the timestamp of the epage tag), make a commit of those resource files in a third git | 12:27.55 |
| | then mash up the epage and new resource repos at the end | 12:28.11 |
| henrys | norbertj: not at all what I expected but I didn't look at it closely. It isn't possible to just have the jpeg decoder return multiple scan lines then just pass them all to the image code? | 12:28.47 |
| Robin_Watts | tor8: That assumes that there were no branches of resources, I think. | 12:28.58 |
| tor8 | by doing subdir filters to move the files (if necessary), grafting the resource commit ids to just before the tag, and filter branch to solidify the grafts | 12:29.02 |
| | of course, that assumes linear history and no tags on branches | 12:29.26 |
| paulgardiner | tor8: one problem is that, since the branch structure of the two modules don't match, we aren't completely sure which versions were intended to be used together | 12:29.27 |
| tor8 | paulgardiner: right, that would make it more problematic to find the right resource set | 12:30.03 |
| henrys | chrisl, kens, Robin_Watts somebody recommend iain banks to me, I assume you guys have read him, any recommendations where to start? | 12:30.11 |
| Jogux | is fairly sure what the intention was, but also pretty certain that what actually happened in practice was different. | 12:30.14 |
| tor8 | actually, scratch that, it should work with the branch structure on the epage git | 12:30.18 |
| Robin_Watts | henrys: oh boy. | 12:30.23 |
| kens | henrys yes we've read Ian *M* Banks | 12:30.25 |
| Robin_Watts | henrys: Iain Banks or Ian M Banks. | 12:30.33 |
| | henrys: Iain Banks or Iain M Banks. | 12:30.39 |
| henrys | the sci-fi M | 12:30.48 |
| Robin_Watts | has read all of both (except The Quarry) | 12:30.50 |
| kens | I liked Excession personally form Iain M Banks | 12:31.01 |
| Robin_Watts | henrys: Same bloke, but the M is used for SciFi books. | 12:31.03 |
| | henrys: It is vitally important that you should read Consider Phlebas first. | 12:31.21 |
| henrys | Robin_Watts: ah that starts a series right? | 12:31.46 |
| chrisl | Hmm, I was going to suggest Player of Games..... | 12:31.46 |
| Robin_Watts | The entire book only works if that's the first one you read. | 12:31.49 |
| tor8 | and the commit history on the new resource git should be irrelevant (all we want is to copy the commits over without history) | 12:32.27 |
| Robin_Watts | He introduces a series with that book, and while that is not the best in the series, you will lose the effect of the book if you read any other one first, IMHO. | 12:32.44 |
| henrys | funny I haven't heard of him I don't think he is as popular here but he has noted success. | 12:32.47 |
| Robin_Watts | henrys: He's massively popular here (well, was, sadly he died a couple of months ago) | 12:33.14 |
| norbertj | henrys: that should also be possible. Normally the jpeg-decoder produces buffers of 8 scanlines. The last buffer will normally not have 8 scans though. Have to think about this a bit. | 12:33.15 |
| paulgardiner | We're hoping that, resources being assets and the like, the history is mainly additions, rather than file alterations, and further hoping that the branches differ mainly in which resource files are included, rather than in the versions of individual files. | 12:33.36 |
| Robin_Watts | I'd advise you to read the books in order. | 12:33.43 |
| paulgardiner | Under those assoumptions, we can create a union of all resource files and use that where we currently have no resource history | 12:34.09 |
| Robin_Watts | I also really like the Iain Banks stuff (he has a wicked and warped imagination) | 12:34.16 |
| kens | lunches | 12:34.43 |
| Robin_Watts | chrisl: Player of Games is one of my favourites too. | 12:34.51 |
| tor8 | speaking of books, I started reading Wool yesterday. I have to admit it is hard to put down the book. | 12:35.04 |
| Robin_Watts | tor.productivity--; | 12:35.19 |
| henrys | Robin_Watts: thanks I'll do that one first getting to the end of Silo, I liked it but as the title suggests - omnibus | 12:35.41 |
| paulgardiner | henrys: You haven't read Iain M Banks? I'm so jealous. I wish I could have the memory of them removed and read them all again. | 12:35.58 |
| norbertj | henrys: though in this case all scans are transferred but there was still 2 bytes left in the input (i.e. the JPEG_EOI bytes). The delayed copying from samples to pw->ptr took care of also reading these 2 bytes I think. Now I immediately copy, but have to take care that the inputbuffer is completely consumed. | 12:36.27 |
| henrys | tor8: I read wool and also liked the prequel silo | 12:37.04 |
| Robin_Watts | henrys: shift ? | 12:37.36 |
| | wool/shift/dirt | 12:37.52 |
| chrisl | Dust | 12:38.07 |
| paulgardiner | is halfway through shift | 12:38.18 |
| henrys | Robin_Watts: sorry yes shift | 12:38.20 |
| Robin_Watts | chrisl: D'Oh. sorry. | 12:38.51 |
| chrisl | I finished Dust just after the staff meeting - thumbs up :-) | 12:38.57 |
| norbertj | henrys: the pxl-testfile I use, has a 30 ReadImages (of 36 scans each), some of which had some scans skipped. And with my first attempt to fix (no delay in copying scans) there was still (at least 1) buffer that had completely decoded all 36 scans but had 2 bytes left in the input. | 12:39.22 |
| Robin_Watts | paulgardiner: Before the git gc --prune --aggressive there were 86K commits. Afterwards 62K. | 12:40.04 |
| paulgardiner | On what repo? | 12:40.48 |
| Robin_Watts | on sot.git | 12:41.50 |
| | I'm just surprised that there were 24K commits that needed --aggressive to get rid of them. | 12:42.22 |
| Jogux | /home/joseph/ChangeLog on casper is the combined cvs history for resources btw, if that's useful. hard to tell which ones where additions and which changes though :-S | 12:43.03 |
| paulgardiner | Not sure what --aggressive does, so not sure how surprised to be | 12:43.04 |
| tor8 | Robin_Watts: git reflog likes to hang on to old commits for a week | 12:43.05 |
| | also run "git repack -ad" to trim the size of the .git to the smallest possible after pruning commits | 12:43.33 |
| Robin_Watts | tor8: This was on a fast-imported converted repo though. | 12:43.36 |
| | tor8: Yeah, I'm calling various commands like that. | 12:44.04 |
| tor8 | Robin_Watts: oh. that's more curious. I thought you had some leftovers from the filter-branch stuff. | 12:44.05 |
| | Robin_Watts: so do we wait for ray to come online to make the dns changes? | 12:44.51 |
| Robin_Watts | I can probaby add mujs.com to the casper setup. | 12:45.17 |
| | but I can't make the mujs.com point to casper. | 12:45.28 |
| | so if you want, I can set you up something so you can start to write pages, and test them. | 12:45.50 |
| | but we'll need ray to make them live. | 12:46.00 |
| tor8 | Robin_Watts: sure. just mirror the mupdf.com approach for the web page? | 12:46.22 |
| Robin_Watts | tor8: exactly. | 12:46.29 |
| | just a mo. | 12:46.37 |
| chrisl | henrys: So, unless there are objections, I'm looking at doing the release tomorrow (although might be Wednesday before it goes live). I'm having to take Friday off, so I'd rather not leave it until later in the week - unless we leave it until next week. | 12:46.54 |
| norbertj | henrys: the jpegdecoder decompresses 8 scans, the pximage.c line 615 now only allocates 1 scan (pxBeginImage:pxenum->row) and would have to know that it is a jpeg image, now it is generic also for rle, deltarow, none: each generating 1 scan at a time. So I think it would make it more difficult in pximage.c to handle this. | 12:48.43 |
| | ops | 12:49.16 |
| henrys | norbertj: i guess xps is broken too. | 12:50.16 |
| | also expects 1 scan line. | 12:50.30 |
| Robin_Watts | tor8, paulgardiner: Ignore what I said about the number of git commits changing. I am a fool. | 12:51.54 |
| paulgardiner | Ah good. It was your turn! | 12:53.13 |
| norbertj | henrys: I'm not sure of that. xps_decode_jpeg has a rbuf = part->data, which is probably the complete data for the image.And the s_DCTD_process is only called once (for the complete image I think).and not per scan as in pximage. | 12:55.10 |
| Robin_Watts | tor8: OK, as far as casper is concerned, it is serving mujs.com | 12:55.30 |
| | the webserver on casper, that is. | 12:55.39 |
| | I've copied the mupdf.com contents into /var/www/mujs.com | 12:55.57 |
| | so you can (should) remove what you don't want and edit the rest. | 12:56.17 |
| norbertj | henrys: but I agree, xps would have to be checked also (as it uses the same s_DCTD_process) | 12:57.00 |
| tor8 | Robin_Watts: fab, thanks. now we just need ray :) | 13:03.31 |
| henrys | norbertj:I'll step through it in the debugger, I'm still not quite understanding how the decode procedure can ever return more than one scan line. | 13:14.12 |
| | norbertj: gs_jpeg_read_scanlines - max_lines == 1 | 13:14.36 |
| norbertj | henrys: but the underlying jpeg-decoder has an internal buffer of 8 scans, that is first filled, and then each call to the jdmainct.c post-processor (color-conversion,. ..) returns 1 scan in samples. | 13:17.21 |
| | henrys: gs_jpeg_read_scanlines is (multiple times) called with inputbuffers, this fills up the internal buffer of 8 scans and if 8 scans are decoded (i.e. jdmainct.c line 356: pmain->buffer_full = TRUE) jdocefct.c:decompress_onepass() does the decoding, and will return JPEG_ROW_COMPLETED (i.e. 8 scans)., gs_jpeg_read_scanlines will return read = 1. | 13:26.04 |
| henrys | norbertj: do you have access to the cluster regressions? | 13:33.45 |
| norbertj | henrys: not yet | 13:33.55 |
| | henrys: it says restricted area | 13:34.57 |
| henrys | norbertj: I can test your patch but it seems to me you should have access | 13:35.03 |
| | norbertj: do you have an NDA with us? | 13:35.25 |
| norbertj | henrys: I think so, but I'll double check with Marten. | 13:35.53 |
| henrys | norbertj: you were using a machine was it peeves? | 13:37.27 |
| norbertj | henrys: yes, but since about a year ago I cannot login anymore ( at least not with FileZilla, which I used to send large testfiles to artifex). | 13:39.05 |
| | henrys: from our side it looks like I connect but then connection is broken of, Perhaps the ssh-keys are not correct anymore? | 13:40.12 |
| henrys | norbertj: I'll try and get your keys from ray when he gets here, then I'll set up a cluster account for you. | 13:41.58 |
| | mace: did you ever touch base with miles? | 13:42.14 |
| Robin_Watts | norbertj: It's more likely to be that casper has changed IP address, and so your side is kicking you out. | 13:42.44 |
| | or peeves, rather. | 13:43.12 |
| | What OS are you on? Windows? | 13:43.21 |
| henrys | Robin_Watts: you can reach peeves? | 13:43.49 |
| norbertj | Robin_Watts: I'm on WIn7 (behind a proxy, but that should work). | 13:43.58 |
| henrys | Robin_Watts: nvm it was just very slow for me. | 13:44.10 |
| Robin_Watts | norbertj: Can you try to connect using putty please? | 13:44.27 |
| henrys | Robin_Watts: the connection is horribly slow | 13:45.12 |
| Robin_Watts | henrys: Yes, I think tor8 might be copying a git repo :) | 13:45.32 |
| norbertj | Robin_Watts: it complains about: Host does not exist | 13:47.55 |
| tor8 | Robin_Watts: henrys: yeah, sorry. It's 2/3 of the way done, should not be more than 45 minutes left... | 13:48.04 |
| Robin_Watts | norbertj: What address are you trying to connect to ? | 13:48.15 |
| | peeves.ghostscript.com ? | 13:48.19 |
| norbertj | Robin_Watts: peeves.ghostscript.com 22 | 13:48.28 |
| Robin_Watts | ping peeves.ghostscript.com and tell me the IP address you see | 13:48.49 |
| norbertj | I can't get through the company firewall | 13:49.25 |
| | Robin_Watts: so it's not resolving into an ip. | 13:49.45 |
| Robin_Watts | Ah. | 13:49.48 |
| | Well, peeves is 184.73.189.105 | 13:50.05 |
| | So you could put that in your hosts file on your PC? | 13:50.20 |
| norbertj | with ping it still times out, so I think the firewall is to thick. | 13:50.39 |
| | Robin_Watts: but with the IP and proxy setup in putty, I get a putty-shell, followed by a Network error: Connection timed out. | 13:51.50 |
| | No prompt or whatever | 13:52.03 |
| Robin_Watts | What IP address? | 13:52.11 |
| norbertj | Robin_Watts: 184.73.189.105:22 | 13:54.00 |
| Robin_Watts | [64.183.45.162] according to my ping | 13:54.35 |
| norbertj | Robin_Watts: if I switch to 64.183.45.162:22 with rlogin (i.s.o. ssh) I get a rlogin username: but no request for passwd (same on 184.73.189.105 b.t.w.). | 13:56.05 |
| | only a Network error: Connection timed out. | 13:56.23 |
| Robin_Watts | norbertj: Are you supplying a key etc? | 13:56.58 |
| norbertj | Robin_Watts: with ssh on: yes | 13:57.20 |
| Robin_Watts | Right. You should use ssh to that IP address with a key. I don't know why this wouldn't be working. | 13:57.52 |
| | There is a /home/norbert | 13:58.38 |
| | tor8: Try: twiki.ghostscript.com ? | 14:05.39 |
| henrys | Robin_Watts: do you have root on peeves - the simple thing to do just copy norbert's .ssh to casper then make an account? | 14:06.28 |
| Robin_Watts | henrys: I do have root on peeves. | 14:06.45 |
| | What are we trying to achieve here? | 14:06.58 |
| | Does norbert want cluster access? | 14:07.10 |
| henrys | Robin_Watts: yes | 14:07.16 |
| Robin_Watts | right. I can sort that out. Give me a few minutes. | 14:07.29 |
| norbertj | henrys: Robin_Watts: it looks like I can't get outside of the company-firewall :( ping of 64.183.45.162 just timesout. | 14:08.33 |
| Robin_Watts | norbertj: Can you ssh to 184.73.189.105 ? | 14:10.00 |
| | oh, right. So it seems the IP address you had before is not peeves, it's casper. | 14:10.20 |
| norbertj | Robin_Watts: 184.73.189.105 = peeves, 64.183.45.162 = casper? | 14:11.06 |
| Robin_Watts | Other way around. | 14:11.18 |
| | Damn. I don't have root on peeves, only peeved. | 14:12.51 |
| | norbertj: Can you email the public key you want to use to robin.watts at artifex.com please? | 14:13.32 |
| norbertj | Robin_Watts:that would be the njj_dsa.pub? | 14:20.02 |
| Robin_Watts | norbertj: I would assume so. | 14:20.38 |
| norbertj | Robin_Watts: sent | 14:21.02 |
| Robin_Watts | norbertj: That's in putty format. | 14:24.45 |
| | I need it in openssl format. | 14:24.49 |
| | It should look like: | 14:24.55 |
| | AAAaaaasdasdsdfdsdf sdfgdfgdfgdf.....== dsa-key-20070928-email-address | 14:25.39 |
| kens | IIRC you need to export it from putty | 14:26.00 |
| Robin_Watts | from putty-keygen I think. | 14:26.13 |
| kens | I think that's correct, I'm just trying to find/remember/reconstruct the sequence | 14:26.34 |
| | puttygen.exe maybe | 14:27.05 |
| | From puttygen.exe, load the private ket, then menu conversions->Export OpenSSH key I believe does it | 14:29.00 |
| | Hmm,or perhaps not | 14:29.53 |
| norbertj | Robin_Watts: my putty-keygen only supports openssh, not openssl format. | 14:33.04 |
| kens | thinks robin means openssh | 14:33.18 |
| Robin_Watts | yeah, opensshm sorry | 14:33.35 |
| | yeah, openssh, sorry | 14:33.40 |
| | (sorry, on phone to garage. My car is gonna cost me a fortune) | 14:33.55 |
| kens | TIme to buy a new one instead ? | 14:34.07 |
| chrisl | Rotor tips gone?? | 14:34.44 |
| norbertj | Robin_Watts: sent (new) keyfile. | 14:40.06 |
| tor8 | Robin_Watts: server not found | 14:41.47 |
| Robin_Watts | chrisl: 4 new tyres (180 each), new coil pack (220), new plugs (120), new plug leads (45), new lambda sensor for the engine (either 130 or 384 depending which one it is). | 14:46.27 |
| | plus labour. | 14:46.33 |
| chrisl | Eek, those are high prices for that stuff..... | 14:47.17 |
| Robin_Watts | Those are, regrettably, bargain basement prices for that stuff :( | 14:47.38 |
| | I can get cheaper types (120 each), but the last set of those only lasted 13000 miles) | 14:48.16 |
| chrisl | Tyres for my porsche I get for ~90ukp each - 225x50x16 Toyo Proxes T1s | 14:49.17 |
| henrys | Robin_Watts: it's a sport car you don't need treads or anything, just slow you down. | 14:49.28 |
| Robin_Watts | http://www.tyrereviews.co.uk/Tyres_For/Mazda/RX8.htm | 14:49.51 |
| | chrisl: ^ | 14:49.53 |
| Jogux | robin_watts : impressive :( | 14:50.03 |
| Robin_Watts | I had Falken 452's on there | 14:50.09 |
| | The Toyo Proxes are interesting. | 14:50.25 |
| | but only 207,160 miles driven, which makes me suspicious. | 14:50.45 |
| Jogux | reckons Robin's repair bill is over double what my car is worth. | 14:50.56 |
| chrisl | They are the best all round tyre I've used | 14:50.57 |
| Robin_Watts | calls garage. | 14:51.04 |
| chrisl | This is my local-ish tyre place: http://www.protyre.co.uk/ | 14:52.01 |
| henrys | Jogux: have you heard from mace? Miles was looking for him. | 14:53.25 |
| Jogux | henrys : he's been busy but I just got a mail from him | 14:54.36 |
| | (just setting up his ssh keys so we can look at ats stuff) | 14:54.45 |
| Robin_Watts | henrys: He was away this weekend, hence didn't reply to miles' mail on friday. | 14:56.04 |
| | chrisl: I need different sized tyres, but even so they are cheaper. 130 rather than 180. | 14:56.37 |
| | So I've gone for that, thanks. | 14:56.42 |
| chrisl | Robin_Watts: what size is on the RX-8, then? I didn't think they'd be larger than the ones I've got | 14:58.09 |
| Robin_Watts | smaller. 45s | 14:58.24 |
| chrisl | Ah, yes, lower profile - I thought they'd be a more common size, which usually brings the price down a bit | 14:59.07 |
| | Robin_Watts: if it helps, I share your pain - I still need to get my car in for new fuel lines, and that's big job on the 924 :-( | 15:02.58 |
| Robin_Watts | chrisl: Yeah, but I only bought a new van 4 weeks ago :( | 15:03.38 |
| chrisl | Robin_Watts: I got rid of my MX-5 ~3 weeks ago...... | 15:04.17 |
| Robin_Watts | chrisl: Ah, want an RX8 to replace it with? :) | 15:04.31 |
| chrisl | No thanks, I'm trying to spend a bit less on car(s) overall, not more ;-) | 15:05.08 |
| Robin_Watts | norbertj: OK, try ssh now? | 15:08.38 |
| Robin_Watts | goes to grab lunch. starving. back in 15 mins. | 15:11.18 |
| mvrhel_laptop | good morning | 15:14.41 |
| kens | Morning Michael | 15:14.47 |
| henrys | hi mvrhel_laptop | 15:14.57 |
| mvrhel_laptop | henrys: so did the meeting date get resolved? | 15:15.11 |
| | need to restart this machine. brb | 15:18.34 |
| norbertj | Robin_Watts: still no success | 15:43.12 |
| | Robin_Watts: I'll have to check with our IT-department | 15:43.45 |
| | Robin_Watts: but do you see any connection attempt from me on peeves? | 15:44.29 |
| Robin_Watts | norbertj: How would I know? | 15:46.04 |
| norbertj | in the logs? | 15:46.13 |
| | Robin_Watts: did you also add my new ssh-key to peeves? replacing the old one. | 15:47.42 |
| Robin_Watts | no. | 15:47.52 |
| | I can't as I don't have root there. | 15:48.03 |
| norbertj | Robin_Watts: ok, so we'll have to wait for Ray? I'm off to home. Bye | 15:50.40 |
| Robin_Watts | norbertj: Sorry, yes. | 15:50.47 |
| norbertj | Robin_Watts: no problem. | 15:50.58 |
| Robin_Watts | tor8: Try twiki.ghostscript.com again? | 15:53.08 |
| kens | I get redirected to 'twiki home' and then asked for a password | 15:54.47 |
| Robin_Watts | henrys: http://www.plosone.org/article/info%3Adoi%2F10.1371%2Fjournal.pone.0088278 | 15:55.08 |
| | kens: Spot on. | 15:55.15 |
| | kens: click through without a password, and you should get the top page. | 15:55.32 |
| | Then you can register. | 15:55.39 |
| kens | Nope, asks me again for password | 15:55.53 |
| | can't ok or cancel it | 15:55.58 |
| | IIf I mae one up it takes me back to the password again | 15:56.45 |
| Jogux | robin_waitts: same here. | 15:56.47 |
| kens | Now cannot close Firefox tab :-P | 15:57.05 |
| Robin_Watts | OK, let me fight it for a bit, sorry. | 15:57.46 |
| kens | NP | 15:57.54 |
| Jogux | kens: hit cancel then ctrl-w really quick :-) (wfm) | 15:59.07 |
| kens | mad clicking did it for me | 15:59.15 |
| Jogux | :-) | 16:00.31 |
| Jogux | suspects Robin_Watts has managed to password protect the "authorization failed" page :-) | 16:00.46 |
| kens | Its a marketing ploy 'now you are here, you can never leave, all your browser belong to us' | 16:01.29 |
| Robin_Watts | Jogux: Yeah, exactly I believe. Try now? | 16:04.34 |
| Jogux | now I get a home page | 16:06.56 |
| tor8 | Robin_Watts: still get 'couldn't resolve host' but I expect that's down to my IPS's dns servers being slow to update | 16:08.13 |
| Robin_Watts | Fab. | 16:08.14 |
| | tor8: yeah, probably not helped by you having tried before. | 16:08.27 |
| tor8 | if I dig @8.8.8.8 I can resolve the hostname | 16:08.28 |
| Jogux | has registered and logged in apparently | 16:08.49 |
| Robin_Watts | We probably ought to create TWiki webs for MuPDF, Ghostscript and GhostDocs. | 16:09.04 |
| | I wonder if I can protect GhostDocs so that it requires you to be an approved user. | 16:09.24 |
| Jogux | would presume so. | 16:10.23 |
| Robin_Watts | Jogux: Looks like it. | 16:11.03 |
| Jogux | though http://twiki.org/cgi-bin/view/Codev/SecuringYourTWiki is not complimentary about that approach. | 16:11.18 |
| Robin_Watts | "Work on the basis that anyone who can edit a TWiki topic could potentially damage your website, and ensure controls are in place to constrain the possible damage they could do" | 16:11.53 |
| | So, I plan to only allow viewing and editing of the GhostDocs stuff to a named group. | 16:12.08 |
| paulgardiner | Sounds best | 16:12.28 |
| Jogux | if it really doesn't preject injection of javascript... boggle. | 16:12.49 |
| | prevent | 16:12.52 |
| Jogux | would imagine it's probably fine for ghostdocs, but I wouldn't go posting customers details on it or similar :) | 16:13.47 |
| kens | I have to head out for an optician's checkup, goodnight all | 16:14.48 |
| Robin_Watts | night kens | 16:15.00 |
| Jogux | marcosw : I've just found one of the ATS images is missing... transferring it from points now; I presume I'm not saturating your line (suspects points's upstream is the limiting factor) but let me know if I am... | 16:29.18 |
| | (missing one is dbserver.img fwiw) | 16:29.36 |
| Henry` | mvrhel_laptop: miles is supposed to send out a trip announcement with the 2nd and 3rd dates, but don't know when. It does work for him. | 16:32.15 |
| | Robin_Watts:difficult to tease out what is correlation there (not causation). There are plenty of happy human populations that range from purely protein and fat (Inuits) to the other extreme. One read of the paper is folks who have been diagnosed with a bad disease might take up being vegetarian. Lots of advise like that around. | 16:38.35 |
| mvrhel_laptop | henry: ok | 16:40.29 |
| | robin_watts: that is an interesting article | 16:42.15 |
| | robin_watts: So I am getting a mupdf crash in fz_open_image_decomp_stream_from_buffer that I did not get before. | 16:50.30 |
| Jogux | hm, is points down? | 16:50.36 |
| mvrhel_laptop | robin_watts is is occurring during fz_realloc_default | 16:51.09 |
| | I will see if I can figure out what is happening | 16:51.27 |
| Robin_Watts | mvrhel_laptop: Hmm. Sounds vaguely like something I might have broken. | 16:51.38 |
| | Is this a particular file? | 16:51.44 |
| mvrhel_laptop | it is the PLRM and occurs during the thumbnail rendering that I am doing | 16:52.09 |
| Henry` | Robin_Watts:I was away from my desk, we you able to square norbertj away with an account, I can follup up with him if you want. | 16:52.27 |
| mvrhel_laptop | so it will take a bit for me to cut it down to something | 16:52.34 |
| | page 662 let me see if I can extract the page and repeat the issue | 16:54.11 |
| Robin_Watts | Henry`: LAst thing we were waiting for him to be able to ssh in. I think he was going to try from home. | 16:54.12 |
| | mvrhel_laptop: Thanks. | 16:54.25 |
| Jogux | ah, points is back again. | 16:56.15 |
| mvrhel_laptop | robin_watts: good news it does it with the single page | 16:56.21 |
| | I am sending it to you | 16:56.24 |
| Robin_Watts | Thanks. | 16:56.41 |
| mvrhel_laptop | it has an image mask in it | 16:56.42 |
| | ok sent | 16:57.59 |
| | I suppose I should have check for additional updates though first | 16:58.35 |
| | robin_watts let me know if you don't see an issue | 16:59.40 |
| | maybe there is something wrong in the winRT streaming stuff that I did | 16:59.52 |
| Robin_Watts | mvrhel_laptop: Hmm. Runs through mudraw without complaint. | 17:06.04 |
| | Let me try a memento build. | 17:06.09 |
| mvrhel_laptop | Robin_Watts: oh actually I am not even using the winRT stuff in this. this is just using the regular windows interface | 17:06.59 |
| Robin_Watts | ok, memento shows memory corruption. | 17:07.22 |
| mvrhel_laptop | robin_watts : ok | 17:07.31 |
| | robin_watts so the call in fz_fill_image_mask is where it starts to go wrong | 17:08.12 |
| | I have another file that is cut down even more | 17:08.23 |
| | let me send that to you | 17:08.26 |
| | you should have that now robin_watts | 17:09.31 |
| Robin_Watts | OK, it goes wrong on fz_close(istm); | 17:09.55 |
| | Oh! | 17:11.22 |
| | istm has already been closed. | 17:11.27 |
| | mvrhel_laptop: In pdf-stream.c in pdf_load_compressed_inline_image, remove the fz_always() { ... } section. | 17:12.14 |
| | Does that solve it for you? | 17:12.17 |
| mvrhel_laptop | let me try | 17:12.24 |
| | robin_watts yes | 17:14.00 |
| | thanks! | 17:14.17 |
| Robin_Watts | I'm just preparing a commit of that now. | 17:14.30 |
| mvrhel_laptop | ok. | 17:14.34 |
| Robin_Watts | mvrhel_laptop: Care to review http://git.ghostscript.com/?p=user/robin/mupdf.git;a=commitdiff;h=2abd6344e9ca31b72fd562fc2ca99cc3fb69445a for me? | 17:15.48 |
| mvrhel_laptop | looks good to me :) | 17:16.02 |
| Robin_Watts | pushed then. Thanks/Sorry. | 17:16.20 |
| mvrhel_laptop | np. glad it was easy to find and not some bone head thing that I was doing here | 17:16.35 |
| | because I have plenty of those | 17:17.00 |
| Robin_Watts | Jogux: You about? | 17:36.32 |
| Jogux | is. | 17:46.33 |
| Robin_Watts | So, we're pondering the idea of NOT fixing the resources in the repo. | 17:52.32 |
| Jogux | nods | 17:52.54 |
| Robin_Watts | And if we ever have to build a tag where the resources are wrong, check out the resources using your cvs hack, and then commit them into the repo and retag then. | 17:53.22 |
| Jogux | given it appears not to be easy, that seems reasonable until such time as we find it does really need to be worried about I guess, and we can always import old copies of resources onto release branches etc if necessary | 17:53.25 |
| Robin_Watts | cool, I think we're in agreement then. | 17:54.24 |
| Jogux | great. so how long will it take? :-) | 17:54.32 |
| Robin_Watts | Well, I think sot.git is ready to go then. | 17:54.52 |
| | paulgardiner is looking at dragging the changes from the old epage tree over into the new one. | 17:55.14 |
| | I'll upload the new version onto casper later tonight, and then we can add the commits later. | 17:55.42 |
| Jogux | oh wait. just to check. Are we going with the "just assume the latest resources is fine for all the old builds" type plan? ie. we'll end up with something like a git repo where the first commit adds the very latest resources? | 17:56.27 |
| paulgardiner | Jogux: assumptions required for that turned out not to be true. | 17:56.59 |
| Robin_Watts | No, we're going with the converted epage+resources cvs -> git | 17:57.12 |
| | so for the trunk, resources will be there. | 17:57.23 |
| Jogux | ah right | 17:57.28 |
| Robin_Watts | but for tags on branches, resources may not be, in which case we will lazily fix them up as required. | 17:57.42 |
| Jogux | so... the release branch me & paul were working on last week won't build then :( | 17:57.45 |
| | but, easily fixed | 17:57.59 |
| paulgardiner | Yeah easily fixed | 17:58.07 |
| Robin_Watts | Jogux: Right. So we do a new commit to fix and retag if required. | 17:58.18 |
| | gotta go. back in an hour or so. sorry! | 17:58.37 |
| Jogux | works for me. | 17:58.37 |
| paulgardiner | Jogux: we can cherry pick (temporarily) your commit to make the build system use the cvs repo on casper. Then the build will pull out the correct version of resources and we can commit it | 18:00.21 |
| Jogux | nods. | 18:00.39 |
| | well. actually. dunno. because it's not running on cvs it may not pick out what it would have done. | 18:01.06 |
| paulgardiner | Ah | 18:01.53 |
| | Good point | 18:02.00 |
| Jogux | but I have to presume it must have picked up the trunk anyway given iirc the branch we were on didn't exist in resources. | 18:02.24 |
| paulgardiner | If we could work out how epage branches map onto resource branches (assuming there is a mapping), we could do a filter branch on that basis | 18:03.53 |
| Jogux | I don't think there is a maaping tbh. my suspicion is that if a customer from a few years back had come back and asked for a new build with just one tweak, they'd have ended up with a build with just that tweak plus the latest resources. | 18:06.11 |
| paulgardiner | If we are fairly confident of that, we can very easily add the latest resources everywhere it doesn't exist | 18:11.13 |
| norbertj_ | Robin_Watts: I tried to connect to peeves from my home-pc and that worked. So there is definitely some work to do on my companies-firewall configuration. | 18:13.08 |
| paulgardiner | There are resource branches that have files in common with trunk but with different contents, although that may be simply because the those branches have not received updates for some time. | 18:13.25 |
| norbertj_ | Robin_Watts: I used the old ssh-key for this. | 18:13.46 |
| Robin_Watts | norbertj_: OK. If you can get past that step, I can then set you up on the cluster and talk you through how to access it. | 18:54.58 |
| Jogux | paulgardiner : yeah, if you look through the changelog for resources there are definitely times that font files were changed to fix bugs. Hopefully any such changes didn't /require/ corresponding changes in the epage tree... | 19:54.27 |
| | Forward 1 day (to 2014/03/25)>>> | |
IRC Logs