Perl 6 - the future is here, just unevenly distributed

IRC log for #darcs, 2014-07-23

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

All times shown according to UTC.

Time Nick Message
00:24 mizu_no_oto joined #darcs
00:27 edwardk joined #darcs
01:03 capicue joined #darcs
01:12 stacey_ joined #darcs
01:16 capicue joined #darcs
01:31 stacey_ joined #darcs
01:45 capicue joined #darcs
05:26 dolio joined #darcs
06:13 lelit joined #darcs
06:34 edwardk joined #darcs
08:02 edwardk joined #darcs
09:21 gh__ joined #darcs
09:44 lelf joined #darcs
09:53 edwardk joined #darcs
09:56 gh__ hi
09:56 lelf` joined #darcs
10:01 gh__ Heffalump: what do you think of http://bugs.darcs.net/patch1177 ?
11:11 mizu_no_oto joined #darcs
11:28 mizu_no_oto joined #darcs
12:08 gh__ joined #darcs
13:43 mdiaz joined #darcs
15:45 mdiaz_ joined #darcs
15:56 gh__ joined #darcs
16:22 Heffalump gh__: I don't have a strong feeling about it
16:23 lelit joined #darcs
16:25 gh__ Heffalump, I'm not sure about the UI either (even if making the log look more like git/hg looks nice to me), but the patch hash can always be changed from A-B-C to C without doing any UI changes
16:28 Heffalump I guess I'd be a bit more concerned about how we define hashes than the precise UI
16:28 Heffalump but I doubt it matters too much
16:36 gh__ both are important in their own way
16:40 gh__ I guess the A-B-C format for patches in OF repos was to have them nicely sorted
17:38 alegadea joined #darcs
17:58 amgarchIn9 joined #darcs
18:28 favonia joined #darcs
18:39 lelf joined #darcs
18:53 favonia joined #darcs
19:42 Heffalump mdiaz_: hi
19:46 mdiaz_ hi Heffalump
19:57 amgarchIn9 joined #darcs
20:02 Heffalump could you go through what you've been doing since last week?
20:04 mdiaz_ No, but I'm doing the blog posts.
20:14 mdiaz joined #darcs
20:20 Heffalump mdiaz: Your last reply was a little confusing - did you mean you're unable to go through what you've been doing, or something else?
20:23 mdiaz Mmm, what do you mean by "go through what you've been doing"?
20:24 Heffalump I meant "list the things you've been doing" (so we can discuss them)
20:25 Heffalump "go through" roughly means "talk about, one-by-one"
20:25 mdiaz I think that if you look at my patches we can discuss something about.
20:26 Heffalump it would be helpful if you could give a high level summary of what you've done, as some of the design principles can be lost in the line-by-line detail of the code
20:29 mdiaz Mmm, what do you need to know about this patch http://bugs.darcs.net/patch1180 for example?  It has several comments on the high-level.
20:32 Heffalump I guess that point more applies to the undo work, then - e.g. what design decisions you've made etc
20:34 Heffalump also, you could talk about anything you tried that didn't work out
20:35 mdiaz Ok, you read this patch http://bugs.darcs.net/patch1182 ?
20:36 Heffalump I've skimmed it, yes
20:38 mdiaz Ok, I think that this patch allow us to undo some commands like amend-record.
20:38 mdiaz But just the last command, for the moment.
20:40 mdiaz Guillaume told me that I can use the function  finalizeRepositoryChanges.
20:41 mdiaz In order to save the state of the repository.
20:41 mdiaz What do you think?
20:42 Heffalump are you asking about whether finalizeRepositoryChanges is the right place to put the code that copies the inventories?
20:42 mdiaz Yes.
20:42 Heffalump I think so
20:43 Heffalump I suspect it belongs more naturally inside the withSignalsBlocked just below, but that's a minor detail
20:45 Heffalump have you thought about any limitations of the current implementation
20:45 Heffalump ?
20:45 mdiaz I still don't understand  this function. Also it is used in many commands, so I don't know if undo will behave well in all commands.
20:45 mdiaz I just tested with amend-record.
20:47 Heffalump have you thought about what kinds of commands it might work with and which ones it might not?
20:48 Heffalump the functio does roughly what its name suggests - it finalizes a set of changes which up till that point are only "tentative"
20:49 mdiaz is this function only  used in commands that don't change the working directory?
20:50 Heffalump I doubt it - it'd be needed for anything that changed the "repository" state
20:50 Heffalump i.e. the committed state
20:50 Heffalump some would change both the working directory and the committed state
20:50 Heffalump but I guess you could look at the code for some commands (e.g. obliterate) to check that
20:51 mdiaz In that case I need to use another function, right?
20:53 Heffalump What do you think should happen in those cases?
20:53 mdiaz Mmm, I don't know. But we just want to undo commands that don't change the working directory.
20:55 Heffalump ok, so do might we need to have two versions of finalizeRepositoryChanges?
20:56 Heffalump s/do might/might/
20:56 mdiaz Or a version of saveDarcsState?
20:58 mdiaz Anyways, that's the best way to "intercept" the commands that we  want to enable for undo?
20:58 mdiaz I mean,  one function on each command?
21:00 Heffalump I don't know of any existing properties of commands that would differentate "Undoable" and "not-undoable" commands.
21:01 Heffalump Also, do we need to do anything when running the not-undoable commands?
21:02 mdiaz Delete the history of commands?
21:04 Heffalump yeah, probably - we can't undo the previous command either any more
21:06 mdiaz Ok, you suggest me to add the property undoable to the commands?
21:07 Heffalump or perhaps a flag to finalizeRepositoryChanges
21:07 Heffalump ?
21:07 Heffalump adding that flag would also let you use the compiler to find all the places that need to be changed
21:07 Heffalump (if you do add a flag, define a custom type like Undoable | NotUndoable, rather than using a boolean flag, as it makes the code easier to read)
21:08 mdiaz Mmm, I think that function will end up with many flags. But ok.
21:09 Heffalump it could always be refactored to take an options record at some point, but i think it'll be ok
21:09 mdiaz Ok.
21:09 mdiaz You know where I can look for the UI?
21:10 Heffalump what do you mean by that?
21:11 mdiaz I was thinking in listing the names of the commands with the date.
21:13 Heffalump I see, to tell the user what they would be undoing? I think that would be very useful.
21:13 Heffalump perhaps that info would also need to be passed to finalizeRepositoryChanges?
21:15 mdiaz Yeah, I think so. Ok, I'll try to make  a prototype for tomorrow afternoon so we can discuss it.
21:16 mdiaz I'll make the blog posts now.
21:16 Heffalump ok, see you same time tomorrow
21:16 Heffalump (roughly)
21:17 mdiaz Ok, bye.
21:43 siel joined #darcs
21:44 favonia joined #darcs
21:55 mizu_no_oto joined #darcs
22:06 siel joined #darcs
23:05 mizu_no_oto joined #darcs
23:59 Rastus_Vernon joined #darcs

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