Perl 6 - the future is here, just unevenly distributed

IRC log for #darcs, 2015-08-17

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

All times shown according to UTC.

Time Nick Message
02:03 favonia joined #darcs
02:03 favonia joined #darcs
02:03 mizu_no_oto joined #darcs
02:14 mempko joined #darcs
03:45 favonia joined #darcs
07:13 favonia joined #darcs
11:00 mizu_no_oto joined #darcs
11:38 maerwald joined #darcs
11:57 maerwald_ joined #darcs
12:08 maerwald joined #darcs
12:19 maerwald joined #darcs
12:47 maerwald joined #darcs
12:54 maerwald joined #darcs
12:56 maerwald joined #darcs
13:58 mempko joined #darcs
14:30 mempko joined #darcs
14:46 mempko joined #darcs
17:54 c74d joined #darcs
17:58 Heffalump notdan: I've figured out what was going wrong with Windows slashes even after your patches; because of the config location move, I didn't have a config file, so it was using the full filesystem paths to identify things, and it doesn't seem to like having a ':' in the path
18:03 Riastradh joined #darcs
18:06 notdan hm, that's strange
18:08 Heffalump when I fixed the config file it is behaving, so it's only the behaviour without a config file that's a problem
18:09 notdan yeah it does have some sort of issues with ":" in file names
18:09 Heffalump they are invalid in many circumstances which may be the reason
18:40 notdan Ok, I think I found the culprit
18:57 Heffalump ready for meeting whenever you are
18:58 notdan Sorry, can you give me 10 more minutes? Meanwhile, I pushed the changes to http://darcs.updog.xyz/updog/darcsden-local (including yet another version of the windows fix)
18:58 Heffalump ok
19:03 notdan OK, I am ready.
19:04 Heffalump cool - how's it going?
19:04 notdan So, what has been done: fixing a couple of bugs in the "record" feature. This includes a nasty-ish when binary files in repo would cause off-by-n errors
19:05 notdan I also made the old dependncy code work and refactored my dependency code to share common code
19:05 notdan Attempt to show media files in an iframe (still WIP)
19:05 notdan "Back" button for the recording thing
19:05 notdan plus some UI tweaks
19:06 notdan overall, I think the record interface is close to being ready
19:06 notdan Oh, and the windows slashes thing : I am not sure if it's fully working tho
19:06 Heffalump I found the next/back thing a bit unintuitive
19:07 notdan Oh :(
19:07 Heffalump ok, the c:\ repos do work now, but the '.' repo is still listed and doesn't work
19:07 notdan oh, yeah, the '.' repo needs a separate fix
19:08 notdan And I am still not sure where did we land on the data race problem?
19:08 Heffalump I haven't got URL launching working properly anywhere still
19:09 Heffalump on Windows it just doesn't make the write call to launch one (I think that may actually be a hard problem) and on Linux it has the timing problem
19:10 Heffalump s/write/right/ # aargh!
19:10 notdan Yeah so that's two separate problems, right?
19:10 Heffalump yes
19:11 Heffalump I just launched den from a directory that wasn't a repo and now I have an invalid repo stuck in the config file that stops it starting
19:12 Heffalump (sorry for the stream of consciousness, just experimenting in random ways)
19:13 notdan Ah, yes, I stumbled upon this issue just a couple of hours ago today
19:13 notdan no, it's good!
19:13 Heffalump another question - what about the security thing to avoid leaking credentials in the process list? I think the plan was a local HTML file with a redirect. It might also help with the Windows URL launching problem.
19:16 notdan Yes, I was hacking on a prototype using javascript and ajax, but there were some problems with cross-origin requests. Reading about it I think I know how to make it work
19:16 Heffalump can't you just use a meta-refresh tag in the HTML?
19:16 notdan As for the Windows URL launch, I found this recently: https://github.com/rightfold/open-browser/blob/master/lib/Web/Browser/Windows.hs
19:16 notdan Haven't tested it yet tho
19:17 notdan Heffalump: uh, that's much simpler and easier...
19:17 notdan Heffalump: i haven't thought of that :(
19:17 Heffalump you're clearly too young or something :-)
19:18 * Heffalump doesn't really know much javascript or ajax so wouldn't even have thought of doing it that way. I should learn more about them.
19:23 notdan so, what do you suggest doing about the back/next buttons?
19:23 Heffalump not sure - I wonder if we need them actually, people can always unrecord or amend-record to fix up mistakes (modulo the lack of UI for that at present :-)
19:24 Heffalump maybe for now "next" can be "preview", that might give the right idea
19:26 Heffalump record needs to be linked from the repo page, right?
19:26 Heffalump still needs, that is
19:27 notdan oh, yeah, preview is a much better name
19:27 notdan Heffalump: yes, we have to add a record button; and *how* is an important decision, imo
19:27 Heffalump I think the long-term goal should be to add unrecord and amend-record and remove the preview page entirely, but for now that sounds good.
19:29 Heffalump 8 of the HTTP tests fail for me btw (with the latest code)
19:30 notdan uh-oh, that's not good :( can you send me the log?
19:30 notdan as for adding buttons, I was thinking of adding another method to the DenInstance class
19:31 notdan and modifying RepoPages.hs to work with those methods
19:31 notdan what do you think?
19:31 Heffalump can you link to the code you mean?
19:34 Heffalump emailed you the tests log
19:40 notdan Well I haven't implemented this, but I was thinnking about having a "additionalLinks :: DDXML" in DarcsDen.Instance.Class and then edit http://hub.darcs.net/simon/darcsden/browse/src/DarcsDen/Pages/PageUtils.hs#122
19:41 notdan But now, thinkinh about this I realize that we should be able not only to toggle the record button, but record functionallity as well
19:41 notdan I wonder if it should be a flag?
19:41 Heffalump what problem are you trying to solve here? Only displaying the record button when we are using the local backend and when the right user is logged in?
19:42 notdan Well I was thinking that the /record handler should be enabled *only* in the local backend
19:42 notdan or not?
19:42 Heffalump I see, so you want to abstract it out properly into the backend definition. Makes sense.
19:43 Heffalump But perhaps it's ok just to have a single boolean that means "include local repository operations", and then define the links themselves in PageUtils.hs or wherever
19:43 Heffalump otherwise it might all end up a bit smeared everywhere
19:46 notdan Yeah, true. But if it's a lot of features under the same flag they have to be smeared everywhere, because they affect different parts of code :)
19:47 notdan But for now I will just add a boolean flag to http://darcs.updog.xyz/updog/darcsden-local/browse/src/DarcsDen/Instance/Class.hs ?
19:47 Heffalump yeah
19:47 notdan I see
19:47 Heffalump I think it's a question of where the code best belongs, and in this case I think code for links is a natural fit in PageUtils.hs or wherever.
19:48 notdan OK, should we add a textarea field (for long comments) the record page?
19:48 Heffalump whereas something that says "define where to find the data for backend X" really does belong with backend X
19:48 Heffalump yes
19:50 notdan I think I might be misunderstanding you, or, rather, we are suggesting the same thing: a Boolean flag and some modification to PageUtils.
19:51 notdan Hm, as for the failing tests -- what happens if you configure darcsden to view those repositories?
19:51 notdan i.e. remove ~/.darcs/darcsden_repos and set the appropriate root in ~/.darcs/darcsden.conf?
19:51 Heffalump yes, we are suggesting the same thing. I'm just re-justifying why it's the "right" approach :-)
19:51 Heffalump which root exactly?
19:53 notdan pointing to the darcsden-local source tree
19:54 Heffalump doesn't seem to make any difference
19:54 notdan But what do you see if you go to the web page?
19:54 Heffalump (in case the influence of that file is deliberate, I should mention that unit tests should be properly isolated :-)
19:54 Heffalump oh I see
19:55 Heffalump I've got '.' as a repo in the list and nothing else
19:57 notdan uh. and there is nothing in ./test-http-data/repos ?
20:01 Heffalump there's stuff there, but why would den find it? Doesn't it need them listed explicitly?
20:01 Heffalump or did you mean I should have launched 'den' from one of those repos?
20:06 notdan oh, sorry, I meant launching darcsden directly
20:06 notdan can you remove the ~/.darcs/darcsden_repo file again tho?
20:07 notdan It should not affect the test, it just affects the listing of repositories
20:08 Heffalump you want me to run 'darcsden' rather than 'den'?
20:08 notdan yes
20:09 notdan because den always tries to add the selected repository to ~/.darcs/darcsden_repos
20:09 Heffalump but then it just goes to CouchDB
20:09 notdan I am just trying to understand what might have caused the tests to fail?
20:09 notdan Heffalump: 'darcsden --local'
20:10 Heffalump ahah
20:10 Heffalump I see the repos
20:10 notdan Hm
20:10 notdan you are using windows?
20:11 Heffalump Linux
20:11 Heffalump I haven't dared to try the tests on Windows :-)
20:11 notdan yeah, I am pretty scared
20:12 notdan Hm, does `cat test-http-data/repos/rep1/_darcs/darcsden/description` do anything?
20:12 Heffalump yes, I get TeST description
20:13 notdan and can you see the description on the web page?
20:13 Heffalump yes
20:13 notdan uh so why do the tests fail :(
20:14 Heffalump want an account on the machine to try?
20:14 notdan oh, snap
20:14 notdan I think it's the HSP version
20:14 Heffalump oh, I may not be using your hacked one any more, I'm not sure..
20:15 notdan can you try the hacked version?
20:15 Heffalump did that have a version bump?
20:15 notdan God, I am sorry for this mess - I still haven't managed to get the parsing to work with the regural hsp
20:15 notdan yes
20:15 notdan to 0.10.0.1
20:26 Heffalump (just working on it)
20:26 Heffalump Can you munge the text after retrieval and before the parsing?
20:27 notdan Theoretically; but last time I tried it was just hard for some reason
20:30 Heffalump you'll need to do something :-)
20:30 Heffalump right, they all pass now
20:30 Heffalump sorry for forgetting about that myself
20:35 notdan OK, so this is a list of issues that I've compiled: https://privatepaste.com/024bc196f5
20:38 Heffalump great, was about to ask for that :-)
20:41 favonia joined #darcs
20:43 Heffalump sounds good
20:46 Heffalump when shall we meet next?
20:47 notdan I would say as soon as you have free time?
20:47 notdan Tomorrow, the day after tomorrow?
20:50 Heffalump tomorrow I can't do until about 9pm London time
20:50 Heffalump Wednesday I can do 8pm but I have a work meeting from 8:30pm-9pm so would need to pause for that
20:53 notdan Hm. Let's say tomorrow, if I have anything good to show. Otherwise Wednesday at 8 or after 9.
20:53 notdan will you be checking IRC tomorrow anyway?
20:54 Heffalump yes
20:54 Heffalump probably be around intermittently before 8 as usual too
21:04 Heffalump see you soon then
21:04 * Heffalump wanders off (still be around off and on for a bit)
22:01 favonia joined #darcs
22:09 mempko joined #darcs
22:20 notdan thanks Heffalump
22:47 mizu_no_oto joined #darcs
23:13 favonia joined #darcs
23:17 mizu_no_oto joined #darcs
23:59 mizu_no_oto joined #darcs

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