Camelia, the Perl 6 bug

IRC log for #darcs, 2013-08-30

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

All times shown according to UTC.

Time Nick Message
00:06 mizu_no_oto joined #darcs
00:34 mizu_no_oto joined #darcs
01:26 mizu_no_oto joined #darcs
01:42 srinup joined #darcs
01:44 intripoon joined #darcs
02:53 mizu_no_oto joined #darcs
02:56 stepcut joined #darcs
03:13 preflex_ joined #darcs
04:06 bsrk joined #darcs
05:03 Heffalump bsrk: hi
05:04 Heffalump bsrk: if I use --from 'ganesh@earth.li' it works
05:05 Heffalump my default author is "Ganesh Sittampalam <ganesh@earth.li>"
05:05 Heffalump So I imagine your logic for detecting the email address is flawed
05:06 bsrk Hi
05:06 Heffalump However if I use --from 'Ganesh Sittampalam <ganesh@earth.li>' I get an internal 'not found' error which is different to the messages I reported yesterday
05:06 bsrk can we postpone for ~15 min
05:06 Heffalump so I don't fully understand
05:06 bsrk I am in another chat
05:06 Heffalump the end time is constrained by my train
05:08 bsrk Okay, let's continue
05:08 Heffalump btw your blogpost is now way overdue
05:09 bsrk The selection of user from email takes the email as a string, there is no parser
05:09 bsrk I will write the blogpost today
05:09 bsrk I guess we will have to add a parser for the email
05:10 Heffalump and make sure this is all tested
05:11 Heffalump btw, your new darcsden code doesn't build on Windows because mime-string depends on iconv
05:11 Heffalump I'm not sure how easy that will be to fix, I've been looking at it a bit.
05:11 Heffalump could you update the todo list, btw?
05:13 bsrk err, what should I update?
05:13 Heffalump ok, it seems that uploading bundles by post completely ignorse the setting of whether there's an "issue tracker" or not
05:14 Heffalump tests for each of the features that aren't tested yet (changse, annotate, bundles)
05:14 Heffalump each of the bugs we know about
05:14 Heffalump breakdown the hakyll support work into smaller pieces so you can cross things off as you go
05:15 bsrk Could you get tests to work on your end?
05:15 bsrk As I recall you were having "prerequisites failed" errror?
05:16 Heffalump haven't managed to yet :-(
05:17 * Heffalump looks at it now
05:17 Heffalump how is hakyll going?
05:18 bsrk I have not done anything, focusing on the next merge for now
05:19 Heffalump what about thinking about security?
05:19 bsrk nothing on that either. :-)
05:20 Heffalump re the merge, which repository is the one sm is supposed to be merging from? I realise I've lost track of the difference between -gsoc and -import
05:21 bsrk He has to import from darcsden-import
05:21 bsrk The name change is simply because I had to juggle with two repos for some time.
05:22 bsrk * right now only darcsden-import exists
05:22 Heffalump I can see darcsden-gsoc on hub
05:22 Heffalump I'm not surprised if he got confused about which one to use.
05:23 bsrk Oh, I got to remove that one
05:23 Heffalump there are things in it that aren't in -import, aren't there?
05:23 bsrk Only patch wise
05:24 arpunk joined #darcs
05:24 Heffalump all the bundle stuff doesn't seem to be in -import
05:24 Heffalump oh, "Add repository bundle feature"
05:25 Heffalump you have two patches named that
05:25 bsrk Yeah, just noticed that
05:26 bsrk Seems like I misnamed one of the bundles
05:30 Heffalump this patch has a typo in the name, btw: http://hub.darcs.net/Aditya/darcsde​n-import/patch/20130829031147-ae621
05:30 bsrk yes
05:31 bsrk Is there anything else?
05:32 Heffalump I'd like to get the todo list updated now
05:32 Heffalump there's quite a lot of stuff we discuss each time that isn't making it there, so it's hard to track your progress
05:36 Heffalump for example there was the annotate UI that we discussed a couple of meetings ago
05:36 bsrk I think I made those changes
05:37 bsrk or I am misremembering what you are referring to.
05:38 Heffalump the UI still looks like the same not very friendly one you started with
05:40 Heffalump we discussed you thinking about alternatives, and/or looking at github, I think
05:40 Heffalump one thing you can do to make sure the todo list is update is to read back over the logs of our meetings to make sure you don't miss things
05:44 bsrk Sorry, just left to collect a package.
05:46 Heffalump there are various things to discuss with sm about features as well as about the merge - for example what he thinks of your new changes from a UI point of view
05:46 Heffalump try to keep those conversations moving along in parallel, and make sure to reply to his emails quickly
05:47 Heffalump what's the story with a couchdb schema upgrade for bundles? Does it need one?
05:48 Heffalump I'm still gettign "Precondition failed" from the tests even once I've got darcsden itself working :-(
05:48 Heffalump can you suggest ways to debug that further?
05:49 bsrk There is a new database and a bunch of views that are new
05:49 Heffalump ok - I have to go, but I think there's quite a lot to be getting on with?
05:49 Heffalump shall we meet same time Monday, given that more of your work is over the weekend now?
05:50 bsrk Ok, I think there are many things to do. I can't do them all,
05:50 bsrk but hopefully the important one's will go through.
05:51 bsrk I'd say that the next merge and send email parser are the top priority.
06:10 mulander joined #darcs
06:59 raichoo joined #darcs
07:32 alexei_ joined #darcs
07:59 lelit joined #darcs
09:02 donri joined #darcs
09:47 owst joined #darcs
10:23 arpunk joined #darcs
11:21 mizu_no_oto joined #darcs
11:31 mizu_no_oto joined #darcs
12:54 bfrank joined #darcs
13:03 favonia joined #darcs
13:33 mizu_no_oto joined #darcs
14:03 int-e joined #darcs
14:03 bfrank__ joined #darcs
15:10 raichoo joined #darcs
15:23 arpunk joined #darcs
15:26 sm morning all.. darcsden feedback complete
15:26 sm (Heffalump)
15:47 mizu_no_oto joined #darcs
15:52 alexei_ joined #darcs
16:29 srinup joined #darcs
16:35 adnap joined #darcs
17:05 Heffalump hi sm
17:17 alexei_ joined #darcs
17:54 mizu_no_oto joined #darcs
18:37 alexei_ joined #darcs
18:48 konundra joined #darcs
20:18 mizu_no_oto joined #darcs
20:23 arpunk joined #darcs
20:37 raichoo joined #darcs
20:59 mizu_no_oto joined #darcs
21:07 haasn cc here: http://hub.darcs.net/ertes/acme-sc​hoenfinkel/browse/Control/Category the Schönfinkelization thing displays incorrectly
21:10 Heffalump serves you right for not calling it Currying
21:10 Heffalump hmm, I think I've seen encoding problems before
21:11 haasn it works for me though http://hub.darcs.net/nand/test-repo
21:11 haasn might have been on his client's side
21:13 haasn Heffalump: (not sure if you're reading the discussion in #haskell) seems to be related to using ssh init
21:13 haasn web interface works correctly
21:17 Heffalump I don't think the way a repo is initialised affects the encoding of recorded patches
21:17 haasn it's the description of the repository, not the patch
21:17 Heffalump oh, I see
21:18 * haasn has a look
21:18 Heffalump so nothing to do with darcs itself
21:23 haasn Heffalump: found the bug
21:23 haasn http://hackage.haskell.org/packages/archi​ve/ssh/0.2.12/doc/html/src/SSH-Util.html
21:24 haasn fromLBS
21:24 haasn map (toEnum . fromIntegral) -- is not an adequate decoding method
21:24 Heffalump thanks!
21:27 preflex_ joined #darcs
21:43 Heffalump sm: your ssh repo doesn't have issue tracking enabled (AFAICT)
21:49 sm Heffalump: hmm, well Alex is still the official maintainer so I was thinking http://darcsden.com/alex/ssh/issues covered it
21:52 Heffalump fair enough
21:52 sm any thoughts alexsuraci ?
21:54 sm haasn noted this (I think) in the darcsden issue tracker also
21:55 Heffalump ah, right
21:55 Heffalump I've just reported it here: http://darcsden.com/alex/ssh/issue/3
21:56 Heffalump perhaps I should just fix it instead of chasing issue trackers :-)
21:56 sm :)
21:57 Heffalump hmm, not entirely ure the problem is there, there's lots of dubious looking conversion code in ssh
21:58 Heffalump well, other places with fromIntegral
21:58 haasn perhaps
21:58 sm indeed, encoding is a notorious rabbit hole
21:58 haasn but I'm quite sure this is it
21:58 Heffalump does the ssh protocol actually have a way to communicate an encoding?
21:58 sm they have tunnels all over the place!
21:58 haasn I looked at where the constructor gets applied and the function eventually boils down to that
21:59 haasn the harder question is “what's correct?”
21:59 haasn just assume utf-8 and move on with life?
21:59 sm sounds good
21:59 Heffalump haasn: could you put the details of where the call that provokes it comes from?
22:00 haasn sure give me a sec
22:00 Heffalump thanks
22:01 Heffalump sm: btw, just a subjective opinion, but I find the issue tracker on darcsden.com less ugly than the hub.darcs.net one
22:02 sm that's interesting, why ?
22:03 haasn Heffalump: http://hackage.haskell.org/packages/ar​chive/ssh/0.2.12/doc/html/src/SSH.html line.. something
22:03 haasn "exec" -> do command <- net readString
22:03 sm brb
22:05 haasn readString comes from SSH.NetReader http://hackage.haskell.org/packages/archive​/ssh/latest/doc/html/src/SSH-NetReader.html
22:05 haasn fmap fromLBS readLBS
22:05 Heffalump sm: I think the main reason is the default sizes, the darcsden one fits nicely into my screen but the hub.darcs one overflows it vertically
22:05 haasn the ‘exec’ thing is what eventually gets passed to our handler in SSHServer.hs
22:06 Heffalump sm: when looking at an existing issue, that is
22:06 Heffalump haasn: great, thaks
22:16 sm Heffalump: oh, is that all ? Can you set a lower size for hub.darcs.net in your browser ?
22:18 sm neither one fits on this screen.. that comment form needs replacing anyway
22:19 Heffalump what do you mean by setting a lower size?
22:20 sm eg press cmd or ctrl - once.. modern browsers will remember that for all pages on the site
22:20 Heffalump but that zooms everything, including text
22:21 Heffalump which may or may not be sensible actually, as the text is a bit big
22:21 sm it sounds like your main issue is not the font size, but the page layout, especially the big comment form ?
22:22 Heffalump right
22:22 Heffalump but ctrl- changes everything
22:23 sm which totally makes sense. Alright, ctrl - isnt your solution then
22:23 sm I think a compact auto-expand-on-focus comment form might be
22:23 Heffalump also, the "revise issue ->" button is just slightly wider than the text entry box for comments, which is visually jarring
22:23 sm like on google's tracker IIRC
22:24 sm sure. like I said, the form needs replacing
22:25 Heffalump fair enough - just thought I'd mention that the form looks ok to me in darcsden.com as what's there may be an alternative in some way
22:26 sm hub's form is almost exactly the one on darcsden.com, just font metrics have thrown it off a bit
22:28 sm and dropping the left button column
22:28 sm drat drat, you force me to fix it
22:33 Heffalump sorry :-)
22:34 Heffalump how come darcsden.com has UI for assigning the bug and hub.darcs doesn't, btw?
22:34 Heffalump (whereas hub does have tags)
22:37 lispy Is darcs bridge suitable for doing one way syncronization? For example, if the upstream is in darcs and you want to view it as a (read-only) git repo, would that work?
22:38 lispy I see a warning on the darcs bridge page: Until further notice, the handling of git-branches is extremely fragile, and almost certain to break. Therefore, I do not recommend darcs-bridge for creating a bridge between multi-head repos, at this point in time.
22:38 lispy But I can't tell from the text
22:38 lispy http://darcs.net/DarcsBridgeUsage
22:38 sm Heffalump: in the interest of simplicity and focus, I disabled assignment for the time being
22:38 sm since tags could cover that
22:39 sm or something like that
22:51 sm ui dev on darcsden is a pain
22:51 sm there's nothing super hard, but it's tedious
22:54 sm joined #darcs
22:55 sm but! there's always the unexpected emacs crash to take ones mind off it!
23:07 mizu_no_oto joined #darcs
23:51 * sm cleans up the issue comment form slightly

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