| <<<Back 1 day (to 2013/03/28) | 2013/03/29 |
onixx | Hello all !!! I have a compilation question I hope somebody could help me with. | 00:42.41 |
Robin_Watts | shoot | 00:43.20 |
onixx | I'm trying to figure if there are configure options I can use to make GS smaller. all I need is ps to pdf (ascii only) | 00:43.22 |
Robin_Watts | onixx: Then yes, you can omit all the devices you don't need. | 00:43.44 |
| ps to pdf means you need the pdfwrite device. | 00:44.11 |
| all the raster ones can go. | 00:44.35 |
onixx | ok , so --with-drivers=LIST ? | 00:44.47 |
Robin_Watts | onixx: Maybe. I don't know the exact details of configure. | 00:45.08 |
onixx | ok, but this has potential to make the binary smaller :-) I will try now ;) | 00:45.30 |
Robin_Watts | The guy you really need to speak to is chrisl, but it's 00:45 in his (our) timezone, so he's not here. | 00:45.31 |
| And he may now be away for the weekend. | 00:45.41 |
onixx | ok, well I can come back next week in here for sure. | 00:45.56 |
| thanks for the tip ! | 00:46.08 |
| I really appreciate | 00:46.55 |
Robin_Watts | you're welcome. | 00:49.41 |
| let me know how you get on. | 00:49.46 |
| ooh. nasty. systemic lcms2 bug. If an allocation fails, it crashes. In at least 96 places :( | 00:51.11 |
| Time to quit. I'll fix it in the morning. | 00:51.36 |
onixx | goo nite | 00:57.18 |
| d | 00:57.20 |
mvrhel_laptop | finally links are working in the windoze app | 03:17.06 |
| hmm an ugly crash though after a few iterations through.... | 03:19.35 |
kens2 | Robin_Watts : ping | 11:28.19 |
Robin_Watts | pong | 11:31.26 |
kens2 | I just did a girt revert when I meant a git reset, then I did a git reset to the commit I wanted, is my repository going to be OK now ? | 11:32.04 |
Robin_Watts | git reset --hard? | 11:33.01 |
kens2 | Umm no I did git reset 'commit' | 11:33.14 |
Robin_Watts | actually... | 11:33.16 |
kens2 | because I wanted to go to the commit marcos was using for valgrind | 11:33.27 |
Robin_Watts | hmm. | 11:33.39 |
| git revert <commit> adds a new commit onto the end of the history that reverts <commit>. | 11:34.09 |
kens2 | Yeah that was my problem, I didn't wantthat (used the wrong command by accident) | 11:34.34 |
Robin_Watts | If you then git reset <commit2> that says "make master be commit2" | 11:34.40 |
| i.e you change the end of your master branch. | 11:34.51 |
kens2 | Yes indeed, that *is* waht I wanted in the first place | 11:34.58 |
Robin_Watts | Thus if you use git reset <previous_commit> that will change your repo back to what it was. | 11:35.15 |
kens2 | So if I later do a git pull rebase then I end up back at HEAD ? | 11:35.22 |
Robin_Watts | But this is NOT what you wanted to do, if I'm understanding you. | 11:35.25 |
kens2 | Nop. I definitely want to discard the unintentional revert | 11:35.40 |
Robin_Watts | Right reset is correct for discarding the unintentional revert, but it's wrong for just changing the commit you're on. | 11:36.14 |
kens2 | Hmm, OK. Let me go back to square one then. git reset --hard ? | 11:36.56 |
Robin_Watts | If you want to jump back in time to a previous position to (say) try to reproduce one of Marcos' problems, then you should use git checkout <commit> | 11:37.03 |
kens2 | AH, OK | 11:37.11 |
Robin_Watts | git reset --hard is a nasty command. If you aren't 100% sure what you are doing, then don't use it. | 11:37.30 |
kens2 | OK well... I need to get my checkout back to HEAD with no changes, how do I do that ? | 11:37.53 |
Robin_Watts | That's one of the few commands that git has that can lose you work with no possibility of rescuing it again. | 11:37.58 |
kens2 | THere is no pending work | 11:38.10 |
| I was clean before I messed up :-) | 11:38.18 |
Robin_Watts | Hmm. some terminology here... HEAD = whatever commit your state currently points at. | 11:38.44 |
kens2 | OK I want to go to the latest commit on master | 11:38.58 |
Robin_Watts | hence if you git reset <commit>, then HEAD becomes commit. | 11:39.01 |
| Well, do git logg -20 | 11:39.20 |
| That should show you where HEAD and master are. | 11:39.35 |
kens2 | ken@ZEN /d/ghostpdl (master) | 11:40.02 |
| $ git logg -20 | 11:40.03 |
| * 8591cc5 (origin/master) pdfwrite - initialise structure member 'stream_md5_val | 11:40.03 |
| | * fee8f49 (cluster/ken) WIP on master: 1ef0083 Memory squeezing bug: lcms2 M | 11:40.03 |
| | |\ | 11:40.03 |
| |/ / | 11:40.03 |
| | * e429ab9 index on master: 1ef0083 Memory squeezing bug: lcms2 MatrixElemTypeF | 11:40.03 |
| |/ | 11:40.04 |
| * 1ef0083 Memory squeezing bug: lcms2 MatrixElemTypeFree | 11:40.04 |
| * 47d384e Memory squeezing fix: gsicc_get_link_profile | 11:40.05 |
| * 3dc5a28 Memory squeezing fix: gsicc_alloc_link | 11:40.05 |
| THat dor tof doesn't look right, but its lost the revert I had before | 11:40.58 |
Robin_Watts | ok, ignore the little "hook" bit of the history with the cluster/ken stuff. | 11:41.23 |
| That's the way the git cluster scripts work. | 11:41.40 |
kens2 | Yeah, its a lilttle confusing but I generally mange to ignore it | 11:41.58 |
Robin_Watts | So I can't see "master" anywhere in that history. presumably it's a long way back. | 11:42.20 |
| So, the first thing we need to do is make master point to origin/master, right ? | 11:42.37 |
kens2 | Well if its back at the commit marcos was using for the valgrind tests it will be a good way back | 11:42.50 |
Robin_Watts | Right. You should never have changed master to get back there, but it's easy to fix. | 11:43.16 |
kens2 | OK | 11:43.23 |
Robin_Watts | Your prompt shows us that you are checked out on master, so to get back to sanity, we just do: git reset --hard 8591cc5 | 11:43.46 |
kens2 | OK HEAD Is now at 8591cc5 | 11:44.19 |
Robin_Watts | That says "Change my current branch (in this case master) so that it actually points to commit 8591cc5, also, because you said --hard, change the index to that too, and throw away any changes in my working set" | 11:44.49 |
kens2 | git logg looks sane now | 11:44.53 |
| So I shoudl do git checkout <commit> to move backwards ? | 11:45.13 |
Robin_Watts | kens2, right, so you should see something like 8591cc5 (HEAD,master,origin/master) | 11:45.15 |
| Perfect. | 11:45.17 |
kens2 | OK let me see | 11:45.35 |
Robin_Watts | That will move HEAD, but not master. | 11:45.36 |
| And you'll go into what they call 'detached HEAD state'. | 11:45.50 |
| (essentially that means that you're not on any named branch) | 11:46.10 |
kens2 | Yes, that's what the message said, thanks Robin_Watts | 11:46.21 |
Robin_Watts | kens2: No worries. | 11:46.44 |
kens2 | presumably I can go back to the master commit by using checkout again ? | 11:46.51 |
Robin_Watts | You can. | 11:46.55 |
kens2 | OK thanks, now to see if I can reproduce the valgrind problem | 11:47.13 |
Robin_Watts | git checkout moves HEAD around (i.e. moves your current state around, and lets you change branch). | 11:47.15 |
paulgardiner | Robin_Watts, tor8: couple of commits up for review. | 13:45.25 |
Robin_Watts | paulgardiner: Ok. will look in just a sec. | 14:06.00 |
paulgardiner | ta | 14:06.09 |
Robin_Watts | paulgardiner: So, printing, eh? | 14:12.15 |
| Where do we actually send the data to the cloud? | 14:12.26 |
paulgardiner | I guess, or maybe direct to the printer. I managed without knowing. :-) | 14:13.07 |
Robin_Watts | paulgardiner: eh? | 14:13.21 |
| I get that this pops up a window to cloudprint. | 14:13.36 |
| but I can't see where we actually say "and here is the file to print" | 14:13.50 |
paulgardiner | There's a javascript callback that accesses the Uri | 14:14.29 |
Robin_Watts | Ah! docUri. | 14:14.42 |
| Is it safe to call fz_meta without fz_try/fz_catch ? | 14:15.51 |
paulgardiner | I believe so. I guess that's something that might change though. | 14:17.39 |
Robin_Watts | ok, then that looks fine. | 14:18.15 |
paulgardiner | great ta | 14:18.21 |
Robin_Watts | I'll push them after lunch. | 14:18.21 |
| paulgardiner: Can you push ? | 14:55.47 |
paulgardiner | I don't think so. I'll give it a try. | 14:56.07 |
| failed to write object. unpack failed | 14:57.02 |
Robin_Watts | git remote -v ? | 14:57.14 |
| I've pushed. | 14:58.04 |
paulgardiner | ta | 14:58.08 |
Robin_Watts | I have 3 reviews on robin/master, but I suspect they may be more tor8ish than paulgardinerish. | 14:58.43 |
paulgardiner | Yeah. Don't really know what the desktop file does. Middle commit looks fine. | 15:02.22 |
Robin_Watts | paulgardiner: I don't know either. It's cribbed from a bug submission. | 15:02.44 |
paulgardiner | Probably best for tor8 to look over the last one. Simple enough and looks right, but I'm not sure I can predict the consequencies | 15:04.36 |
Robin_Watts | paulgardiner: Yeah. It looks right to me too (and I wrote the image decode shortstopping in the first place). Cluster tests show it's fine. But presumably I had a reason for not doing indexed ones in the first place. | 15:07.30 |
paulgardiner | One more little commit up for review. Would be handy if I could push. No wish to push without review, but it would save you the hassle of branch juggling when you're in the middle of a change. | 15:15.28 |
Robin_Watts | paulgardiner: Yeah. I'm not entirely sure how it can be set up so I can push but you can't... | 15:15.54 |
paulgardiner | I think it is possible for you to push mine without disturbing yours but it's a tortuous sequence of commands (or at least the only way I can think of) | 15:18.31 |
Robin_Watts | paulgardiner: I can git fetch paul master then git push golden hash:master | 15:19.28 |
| so no problems there. | 15:19.33 |
| I think you're a member of gs-priv, but not gs-pub. | 15:19.49 |
| Which seems oddly the wrong way around :0 | 15:19.57 |
paulgardiner | What's hash:master ? | 15:20.48 |
Robin_Watts | git push <repo> <what it's called on my end>:<what I want it to be called on the other end> | 15:21.44 |
paulgardiner | Ah handy | 15:22.16 |
Robin_Watts | if <what it's called on my end> == <what I want it to be called on the other end> then you can omit the : and one of them. | 15:22.18 |
| hence git push golden master~2:master will push all but the last 2 commits on my master up to the other end. | 15:22.47 |
| and git push golden :master will delete master on the remote. | 15:23.00 |
paulgardiner | I knew the delete one, but not that it was a specific case of something more general | 15:23.32 |
| So you'd do git push golden paul/master:master ? Or do you need to assign a local branch? | 15:24.05 |
Robin_Watts | paulgardiner: I'd just copy the SHA from the web view :) | 15:24.33 |
paulgardiner | Ha right :-) | 15:24.41 |
Robin_Watts | If you want to create a new branch on the remote, then you need to push from a local branch, otherwise it doesn't know whether you are making a branch or a tag or what. | 15:34.07 |
| but otherwise it's pretty smart. | 15:34.17 |
| We've got a few files that regularly SEGV in mupdf in the mujstest stuff. | 15:49.03 |
| I think it's store related, but I can't reproduce it under conditions where I can attempt to debug it :( | 15:49.26 |
| Morning mvrhel_laptop | 16:04.22 |
mvrhel_laptop | good morning | 16:05.11 |
henrys | does not like: c-style.htm Don't spend a lot of time commenting executable code unless something unusual or subtle is going on. | 16:09.04 |
| I don't think we should encourage few comments in ghostscript | 16:09.34 |
Robin_Watts | Code which has comments for every "blah++" is bad. Code which has no comments at all is bad. ghostscript is towards one end of that spectrum :) | 16:11.40 |
kens | I break c-style all teh time, because I put in lots of comments | 16:18.34 |
| Otherwise I cannot remember what my code does.... | 16:18.52 |
mvrhel_laptop | yes. I am an over commenter too | 16:27.04 |
| due to my short memory span | 16:27.15 |
Robin_Watts | mvrhel_laptop: That's not over commenting then :) | 16:28.04 |
mvrhel_laptop | I suppose it is all relative | 16:28.17 |
Robin_Watts | (It might be according to c-style, but...) | 16:28.25 |
mvrhel_laptop | have my links working crash free in the windows app now. need to add in the contents stuff, reflow, and copy/paste then I think that will be enough | 16:28.51 |
| no past, just copy i mean | 16:29.18 |
Robin_Watts | The trick is to leave code in a state whereby when you come back to it a year later without having touched it since, there are no "WTF?" moments. One day I'll manage that. | 16:29.39 |
mvrhel_laptop | :) | 16:29.55 |
henrys | Robin_Watts: game of thrones sunday nightâ¦. things were getting bad I was watching Duck Dynasty | 16:37.52 |
Robin_Watts | henrys: Yeah, we just rewatched season 2 in preparation. | 16:38.11 |
| We're mid way thrugh S5 of SOA. I see what you mean about the violence ratcheting up. | 16:38.30 |
| Are you watching "The Following" ? | 16:38.59 |
henrys | no I thought I'd wait for net flix to have it. | 16:40.02 |
| do you like it? | 16:40.13 |
Robin_Watts | yes, more than I expected. | 16:40.20 |
| Revolution starts here tonight. | 16:40.58 |
henrys | usually not a big Bacon fan | 16:40.59 |
| I know you are ;-) | 16:41.16 |
Robin_Watts | haha! | 16:41.20 |
frobert | hi there | 17:12.34 |
ghostbot | moin moin | 17:12.34 |
frobert | I am looking for some advice on TrueType font embedding | 17:13.05 |
| I have code that generate Type 42 from TTF files | 17:13.58 |
| But Ghostscript 9.06 says Warning: 'loca' length 3418 is greater than numGlyphs 242 in the font AAASUB+ArialMT. | 17:15.10 |
| In fact, it always complain that numGlyphs is 242. | 17:15.32 |
| Which I suspect is because the maxp TTF table is not read properly | 17:15.56 |
kens | maxp is irrelvant as that is the maxcimum number of poitn IIRC | 17:16.12 |
| its sayign that the declared loca length is too large for the declared numGlyphs | 17:16.31 |
frobert | The maxp table starts with 0001000009E600F2... | 17:16.41 |
| F2 = 242 | 17:16.46 |
| I think there is a byte order issue when reading the maxp | 17:17.05 |
| I suspect GS looks at 00F2 to be numGlyphs, whereas it should look at 09E6 | 17:17.45 |
| Hi ken | 17:18.05 |
kens | I would need to see the font | 17:18.13 |
| And I'm about to log off for a long weekend | 17:18.29 |
frobert | Font is stock Win7 Arial (or Times) | 17:18.56 |
kens | Can you either come back next week or raise a bug report on http://bugs.ghostscript.com please and attach the font there. | 17:19.05 |
| You say you are making Type42 from TTF, its the type 42 I need to see | 17:19.20 |
frobert | I swapped the bytes in the sfnts arrray and guess what ? GS stopped complaining ! | 17:19.23 |
kens | Like I said, I need to see the font | 17:19.33 |
frobert | Do you need the T42 code as well ? | 17:19.58 |
kens | frobert : no just what you are feeding into Ghostscript | 17:20.19 |
frobert | ok | 17:20.29 |
| I'll raise a bug report then | 17:20.44 |
kens | OK it wil be next week before someone can look into it though | 17:20.59 |
| In the meantime, goodnight all | 17:21.50 |
frobert | Ciao | 17:21.55 |
ray_laptop | gnite, ken | 17:22.00 |
Robin_Watts | have a good one kens | 17:22.04 |
| bah. | 17:22.06 |
ray_laptop | Robin_Watts: that bug for cust 532 turned out to be a manifestation of bug 693611 that chrisl just fixed on Mar 19th. | 17:22.47 |
Robin_Watts | yeah, I just read the email. Nice. | 17:23.02 |
ray_laptop | strange that it should show up so closely in time. It's been in their code AND our HEAD code for a LONG time! | 17:23.47 |
| and it isn't really that rare -- having to do with when a GC runs when we are doing transparency. Maybe we don't RELOC the structures very often in the normal case as was happening for the pattern clist | 17:25.34 |
Robin_Watts | ray_laptop: Could you install cgdb on peeves please? | 17:26.27 |
| (I assume it's just an apt-get install cgdb) | 17:27.00 |
henrys | Robin_Watts: emacs -nw M-x gdb ⦠same thing | 17:29.37 |
Robin_Watts | henrys: oooh. I will try that right now. | 17:30.11 |
ray_laptop | Robin_Watts: OK. just a sec (phone call...) | 17:31.54 |
Robin_Watts | henrys: How is that not just gdb running in a shell? | 17:32.32 |
henrys | it bring up a window tracing where you are in the source. | 17:33.13 |
Robin_Watts | Not for me :( | 17:33.19 |
henrys | did you start a shell and run gdb? | 17:33.35 |
Robin_Watts | emacs -nw | 17:33.45 |
| then ESC x gdb | 17:33.52 |
henrys | maybe something weird on peeves let me look | 17:34.17 |
Robin_Watts | Ah, now it works. | 17:35.18 |
ray_laptop | Robin_Watts: the install fails :-( | 17:35.26 |
Robin_Watts | ray_laptop: Don't worry - I'll try emacs for now. Having the source displayed should make this a much nicer task. | 17:35.53 |
| Thanks for trying. | 17:35.58 |
| henrys: Thanks for the emacs pointer. | 17:36.04 |
henrys | np | 17:36.13 |
ray_laptop | Robin_Watts: BTW, the error I get is: Failed to fetch http://us.archive.ubuntu.com/ubuntu/pool/universe/c/cgdb/cgdb_0.6.4-3_amd64.deb 404 Not Found | 17:38.52 |
| I suspect that it is because my ubuntu is so old (and I'm hesitant to update it) | 17:39.30 |
| I'm running 2.6.31-23-generic #75-Ubuntu SMP | 17:40.24 |
Robin_Watts | ray_laptop: I understand completely. I find that either you take every upgrade that is offered to you when it's offered to you, or stuff falls out of date. | 17:40.25 |
ray_laptop | tkamppeter: so how safe is updating to a newer ubuntu base ? The particular issue is getting the init carried forward for my md RAID array | 17:42.04 |
Robin_Watts | tkamppeter: I have upgraded from one version to the next just fine in the past -but then I am running in a VMware instance, not on real hardware. | 17:48.09 |
| ray_laptop: Have you considered buying a new harddrive, cloning, then upgrading the clone ? | 17:48.27 |
| HDs are cheaper than your time, right? | 17:48.39 |
| paulgardiner has 2 servers and flip/flops them when he upgrades versions. | 17:49.10 |
| oh, raid. Multiple HDs :( | 18:16.42 |
| Buy an SSD, put the base OS on that, mount the raid stuff on that for bulk storage? That way you can still boot even if the raid stuff doesn't work immediately ? | 18:17.58 |
| Are you raiding for speed or safety? or both? | 18:18.25 |
ray_laptop | Robin_Watts: I already have an SSD for the OS and /tmp and /dev/mdo is where /home lives (the rest). I run RAID-5 with 3 disks for speed and security | 18:24.53 |
Robin_Watts | ray_laptop: So, buy a new SSD, clone onto it, upgrade that, and you can back out if required? | 18:32.04 |
| or clone the SSD onto an old HD and get that working as your boot drive? Then when that works clone back onto the SSD? | 18:32.51 |
ray_laptop | Robin_Watts: I like the first approach. Get a new, faster, SSD. This one is fairly old and slow by today's standards | 18:52.46 |
| I might follow marcosw's suggestion of a card based SSD rather than the sata (I think henrys did that) | 18:54.14 |
Robin_Watts | ray_laptop: yeah. Its more expensive, but much faster, AIUI. | 18:56.07 |
ray_laptop | Robin_Watts: you sure are keeping busy on the memsqueeze fixes. Nice to get these addressed | 18:56.15 |
Robin_Watts | OUCH! http://www.tomshardware.com/reviews/pci-express-ssd,2952-18.html | 18:57.38 |
| $700 for 240Gig, and that for the cheaper ones. | 18:58.01 |
ray_laptop | Robin_Watts: do you think Miles would go for this one: http://www.serversupply.com/SSD/PCI-E/3.2TB/OCZ%20TECHNOLOGY/ZD4CM88-FH-3.2T.htm | 18:58.23 |
Robin_Watts | ray_laptop: You'd want a pair of them for redundancy. | 18:59.08 |
ray_laptop | the OCZ RevoDrive 3 PCI-Express SSD solid state drive - 120 GB is probably good enough. It's only $319 | 19:01.29 |
Robin_Watts | That's a factor of 2 or so more than SATA based ones. That's more reasonable. | 19:02.28 |
ray_laptop | but 3.2 Tb of SSD _would_ speed some things up. 2800 MBPS read and write speed | 19:02.33 |
Robin_Watts | 2.8GBPS :) | 19:02.57 |
ray_laptop | that's about 15x over my current SSD. Most today are in the 300-400 MBPS | 19:05.01 |
Robin_Watts | my new one is 540/250 (read/write) | 19:06.06 |
ray_laptop | Robin_Watts: I notice marcosw is assigning some of the clist (cmd_ ...) valgrind issues to you. Thanks :-) | 19:06.40 |
Robin_Watts | yeah. I will have words with him :) | 19:06.57 |
ray_laptop | OK, now to shift gears and look into the issue for cust 801 | 19:16.54 |
| eventually I will get back to the pdf14 using clist as needed for non-clist devices :-/ | 19:17.48 |
| bbiaw | 19:18.03 |
mvrhel_laptop | ok. external uri links work in the windows app now too, launching the browser. | 20:56.12 |
| now onto the contents stuff | 20:57.08 |
| bbiab | 20:57.12 |
ray_laptop | mvrhel_laptop: I looked at your write-up. I had of couple of (very non-urgent) questions, so next week is fine | 21:03.30 |
Robin_Watts | anyone understand anything about psdf_DCT_put_params etc? | 21:08.57 |
mvrhel_laptop | ray_laptop: I sent an updated write up to you | 22:52.22 |
| ok. it is sunny and 65 here in seattle. time to end the day | 22:52.36 |
| Forward 1 day (to 2013/03/30)>>> | |