Camelia, the Perl 6 bug

IRC log for #cdk, 2010-09-10

| Channels | #cdk index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
03:26 azeem left #cdk
03:27 azeem joined #cdk
05:35 egonw joined #cdk
06:27 jbrefort joined #cdk
07:14 Gpox joined #cdk
08:07 egonw_ joined #cdk
08:10 egonw left #cdk
09:03 egonw_ is now known as egonw
09:49 zarah joined #cdk
09:58 bag_ joined #cdk
11:18 maclean joined #cdk
11:25 egonw hi maclean
11:25 maclean hi egonw
11:27 maclean oh, the concrete task is just this : "Fix Fonts" :)
11:27 egonw well...
11:27 egonw I like to hold that off until the zooming is fixed
11:28 egonw and scaling
11:28 maclean ugh.
11:28 egonw yes
11:28 egonw (or, if it really is an easy fix...)
11:28 maclean fixed like minor changes or fixed like completely reworked?
11:29 egonw the current zooming is done wrong...
11:29 egonw call it reworked...
11:29 egonw accurate term would be otherwise
11:29 egonw as the current impl is against the original design
11:29 maclean sigh.
11:29 * egonw agrees and is annoyed that he has not noted this much earlier
11:30 egonw and jumped in at a more appropriate time :(
11:30 maclean well, I made a toy implementation of the architecture yesterday.
11:30 maclean and it seemed to be fine.
11:30 maclean Of the existing architecture, that is.
11:31 maclean I can use it to test out changes, to check that it all works.
11:31 maclean Of course, (unit) tests will be more useful in the long run.
11:32 egonw tell me more about this toy impl
11:32 maclean It only renders 'textboxes' and lines between them.
11:32 maclean But it uses a Model, a Renderer, aaaandd...
11:33 maclean A Diagram composed of Elements.
11:33 maclean Well, we don't have an explicit diagram class in the real architecture, but still.
11:33 maclean Oh, and of course, a FontManager :)
11:34 egonw I guess the design should be like this:
11:34 egonw drawing of a text element is done in world coordinates
11:34 egonw but since we do not know who large this will be in pixels... no chatting about fonts yet
11:34 egonw (not fully aware of the current impl)
11:35 egonw this text boxes can zoomed/scaled...
11:35 maclean Well, the center point of the text can be determined in both model and screen coords.
11:35 maclean Sorry, I forget which 'world' coords refer to.
11:35 egonw actually... shouldn't the affine transformations take care of font scaling?
11:35 egonw world vs. screen
11:36 egonw screen is pixels...
11:36 egonw world is angstrom
11:36 maclean right, so world == model.
11:36 maclean (for me :)
11:37 maclean Affine transforms don't scale the fonts, no.
11:37 maclean Or at least they don't in my toy.
11:37 maclean Originally, there was an OSX bug that made this impossible.
11:37 egonw oh, yes, the lovely OS/X
11:37 egonw apple--
11:38 maclean I agree that it makes sense to use the underlying graphics code to do as much work as possible
11:38 maclean But so long as this doesn't prevent us doing stuff that would be useful.
11:38 egonw wel, there is now a lot of broken code where coordinates are zoomed but not scaled
11:38 egonw which is probably the reason why the margin is broken
11:39 egonw the problem right now is this useful
11:39 egonw us developers have the tendency to define our desktops as defining the usefulness
11:39 maclean Zoom and scale are both in the transform, so it's weird that the margin is broken.
11:40 egonw zoom shows up all over in the rendering code
11:40 maclean you haven't seen the "extraZoom" parameter in the JCP code then? :)
11:40 egonw oh no...
11:41 egonw what??
11:41 egonw say again??
11:41 egonw no, don't tell me
11:41 maclean Hahaha! Not my fault :)
11:41 egonw I do not want to know about that
11:41 maclean Fair enough.
11:41 egonw I really hate this concept: "oh, we have a flat tire... let's add a fifth wheel to the car"
11:42 egonw damn useful
11:42 egonw as you can drive the car again.... but .... but ...
11:42 maclean True, but remember when making changes that there are 3 main 'clients' (I no particular order) a) JCP b) Bioclipse/SWT c) static images.
11:44 maclean Where c) is often neglected - it involves fitting to canvas, for example. Also margins are important. Oh, and (apparently) there is this requirement to scale to 1000%...
11:44 maclean Anyway, my plan would be to try to make test cases that cover these use-cases, so that changes that improve one don't degrade others.
11:46 egonw c) is prime part of the groovy cdk-jchempaint scripts too
11:47 maclean Ah good.
11:47 maclean Asad is also keen to get static images to look nice.
11:48 maclean But rendering in general, of course.
11:49 egonw good
12:01 * sneumann is also keen on static images
12:02 maclean Many people are. I think there is a definite need for a class that just 'does' molecule images.
12:02 maclean Like "Image image = MoleculeImage.makeImage(myMolecule);"
12:03 maclean where all the nasty stuff is hidden away :)
12:04 maclean In fact, I remember a guy at the EBI coming to find me to ask about this functionality as it used to exist, but no longer...
12:04 egonw doesn't rajarshi have such code?
12:05 maclean Probably.
12:08 egonw damn, his server is still down
12:08 egonw well, check later: http://rguha.net/code/java/
12:46 egonw maclean: more RDF stuff in my blog: http://chem-bla-ics.blogspot.com/2010/09/pu​lling-out-data-as-json-from-xhtmlrdfa.html
12:46 zarah egonw's link is also http://tinyurl.com/392fz4r
12:46 maclean ok
12:49 sneumann worksforme: http://rguha.net/code/java/
12:50 maclean egonw: ack
12:50 maclean sneumann : yes, seems to be back.
12:51 egonw I guess Rajarshi is now awake then :)
12:53 egonw indeed: http://friendfeed.com/rajarshi/5​d65ae4f/pig-and-cheminformatics
12:53 egonw (see Rajarshi's comment :)
12:57 maclean egonw: should I replace the patches with their signed-off versions, or append?
12:59 egonw replace sounds good... use the comment field to mention it is the signed of version
12:59 maclean ok
12:59 egonw perhaps keep the original in
12:59 egonw yeah, that might be better
12:59 maclean ok
12:59 maclean this process should really be smoother...
13:00 egonw agreed
13:02 maclean I should really make the distributed reviewing client that I was planning (called 'Cricket') :)
13:03 egonw yeah, would be great...
13:03 egonw have you seen gerrit, btw?
13:03 maclean no?
13:04 egonw I have not had time to look at it...
13:04 maclean (will google)
13:04 egonw only heard about it
13:04 egonw https://code.google.com/p/gerrit/
13:05 maclean my vague sketchy plan would be a program that alternates between editing and reviewing of any document - using external editors, of course.
13:06 maclean So you could act in 'writer' mode, and make repository branches (git, hg, etc) or switch to 'reviewer' mode and get/select/test patches.
13:08 egonw sounds like a lot of work too
13:08 egonw (on top of 'very interesting')
13:08 azeem the vmware people wrote something as well I believe
13:08 maclean heh :) that's why it's just sketchy drawings in my notebook.
13:09 azeem ah right, Review Board
13:09 azeem http://www.reviewboard.org/
13:09 maclean hmmm. looks interesting.
13:10 maclean but I was thinking more of a desktop, not a web solution. perhaps that's outdated thinking.
13:11 azeem the web is the new desktop
13:11 maclean :)
13:11 egonw is there a difference??
13:11 maclean egonw: offline use?
13:13 azeem what do you mean, offline?
13:13 maclean heh.
13:14 maclean Naturally, I am always jacked-in to the matrix.
13:16 sneumann Google Mondrian: web-based code review and storage
13:16 sneumann http://www.niallkennedy.com/blo​g/2006/11/google-mondrian.html
13:16 sneumann Rietveld: a new code review tool
13:16 sneumann http://lwn.net/Articles/281031/
13:16 sneumann have a nice weekend, bye
13:16 maclean bye
13:18 egonw bye
13:20 sneumann left #cdk
13:22 CIA-98 cdk: Egon Willighagen cdk-1.2.x * r9c1b95a / src/test/org/openscience/cdk/atom​type/CDKAtomTypeMatcherTest.java :
13:22 CIA-98 cdk: Unit test to reproduce failing atom type perception with one of the options to create a -1 Integer object
13:22 CIA-98 cdk: Signed-off-by: maclean <gilleain.torrance@gmail.com> - http://bit.ly/8XkjS5
13:22 CIA-98 cdk: Egon Willighagen cdk-1.2.x * r324f7f5 / src/main/org/openscience/cdk/at​omtype/CDKAtomTypeMatcher.java :
13:22 CIA-98 cdk: Compare values not objects (fixes #3061263)
13:22 CIA-98 cdk: Signed-off-by: maclean <gilleain.torrance@gmail.com> - http://bit.ly/dkxpoz
13:23 maclean egonw: a small point, but I assumed you wanted me to mark the patch as 'accepted' when I uploaded the signed-off patches?
13:23 egonw umm... that's the point if signing off...
13:23 egonw indicated that you accept the patch
13:23 maclean sure, ok.
13:23 maclean that's what I assumed
13:24 egonw ok :)
13:24 egonw thanx
13:25 egonw now I can release 1.2.7 this weekend
14:11 maclean left #cdk
14:14 maclean joined #cdk
14:54 maclean left #cdk
16:01 egonw left #cdk
16:01 egonw_ joined #cdk
16:05 egonw_ left #cdk
16:49 slyrus joined #cdk
17:21 Gpox left #cdk
18:03 slyrus left #cdk
19:31 egonw joined #cdk
19:58 egonw_ joined #cdk
19:58 egonw left #cdk
21:08 jbrefort left #cdk
21:46 egonw_ left #cdk
22:28 slyrus joined #cdk
23:14 bag_ left #cdk

| Channels | #cdk index | Today | | Search | Google Search | Plain-Text | summary