Perl 6 - the future is here, just unevenly distributed

IRC log for #darcs, 2017-03-18

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

All times shown according to UTC.

Time Nick Message
00:45 Riastradh joined #darcs
02:48 ilbot3 joined #darcs
02:48 Topic for #darcs is now http://darcs.net/ | logs: http://irclog.perlgeek.de/darcs/ | darcs 2.12.5 is out http://darcs.net/Releases/2.12
05:08 mizu_no_oto joined #darcs
05:33 favonia joined #darcs
05:45 pointfree[m] joined #darcs
08:15 sm @tell jeltsch what's an example ?
08:15 lambdabot Consider it noted.
08:55 fabricandink joined #darcs
09:01 fabricandink Hi Heffalump! I tried every combination of quotes and double-quotes.
09:01 fabricandink (sorry for the nick-name confusion)
09:02 fabricandink dar pull --external-merge "TortoiseGitMerge.exe /base:%a /mine:%1 /theirs:%2 /merge:%o" => Unexpected 'a'
09:02 fabricandink dar pull --external-merge "TortoiseGitMerge.exe /base:"%a" /mine:%1 /theirs:%2 /merge:%o" => unexpected ' " '
09:03 fabricandink the outer quotes don't seem to make any difference. i tried '...' and "...".
09:12 Heffalump hi
09:12 Heffalump one sec, just want to read through the docs again
09:15 Heffalump (this is not being assisted by darcs.net apparently being down..)
09:31 Heffalump darcs pull --external-merge 'TortoiseGitMerge.exe "/base:%a" "/mine:%1" "/theirs:%2" "/merge:%o"'
09:31 Heffalump can you try that?
10:38 fabricandink joined #darcs
10:48 fabricandink joined #darcs
10:52 fabricandink_ joined #darcs
10:52 fabricandink_ Heffalump: that works!
10:53 fabricandink_ at least, tortoisemerge is opening
10:53 fabricandink_ but it doesn't like the path
10:53 fabricandink_ >_<
10:54 fabricandink_ bad syntax: 'appdata\local\temp\ancestor\.\hello.txt'
10:56 fabricandink_ jesus...
11:13 Heffalump hmm, that's unfortunate..
11:13 Heffalump do you think it's the path with .s in that it doesn't like?
11:41 bfrk joined #darcs
11:42 bfrk Hi
11:42 Heffalump hi
11:43 bfrk I have a question about darcshub: is there a way to manage multiple branches of a project under a common project name?
11:45 bfrk I am considering to move one of my projects there but I have to maintain branches for a few versions
11:47 Heffalump no - I just use suffixes
11:47 Heffalump I think it'd be nice to have a subdirectory structure there
11:50 bfrk just an idea: perhaps we should add minimal support for sub-repos in darcs
11:59 Heffalump bfrk: in-repo branching in some form?
11:59 Heffalump my current idea (though I haven't really worked on it for a while) is to be able to suspend and unsuspend patches, and then use tags to track branches
12:02 bluepixel that would be awesome. :)
12:07 sm hi bfrk, I use REPO_branchname
12:07 sm REPO_BRANCHNAME
12:21 bfrk joined #darcs
12:23 bfrk Heffalump: i was thinking about something like git submodules
12:33 pointfree[m] joined #darcs
12:42 jeltsch joined #darcs
13:28 pointfree[m] http://darcs.net/Ideas/Stash
13:37 pointfree[m] I like to think of it as twitter hashtag aggregation for darcs
13:45 pointfree[m] That is why, IMO, "darcs stash" should really be called "darcs tag". But "darcs tag" is taken by another subcommand.
13:47 bfrk pointfree: i can't say i understand, how is stash and tag similar?
13:48 Heffalump pointfree[m]: darcs stash is what I mean about suspending above
13:48 Heffalump it works well enough that I've been using it locally for a while, but it's nowhere near release-ready
13:49 bfrk i thought stash is like rebase suspend but always keeps patch identity
13:51 bfrk Heffalump: can i try it?
13:59 pointfree[m] The weak hash feature and "darcs tag" are special cases of selecting a subset of patches. Maybe the definition of a TAG should be expanded to include patch subsets not going all the way back to the origin.
14:00 Heffalump http://hub.darcs.net/ganesh/darcs-stash-draft-20150914
14:01 Heffalump and yes, "stash" suspend is like rebase suspend but keeping identity
14:01 Heffalump note the "stash" name is just temporary, the UI we (a few us at the last Paris sprint) discussed was to just be "darcs suspend" and "darcs unsuspend"
14:02 Heffalump I think "darcs tag" is correctly used to label repo states
14:02 Heffalump but I want to be able to use suspend/unsuspend to *move* to a repo state (non-destructively)
14:05 pointfree[m] but what if you want to "tag" a set of patches to suspend and unsuspend -- like bookmarks.
14:05 bfrk pointfree[m]: understood; but i think i wouldn'use stash for that
14:08 Heffalump yes, that's another important case, being able to select a group of patches, not just a repo state
14:08 Heffalump I had in mind two tags for that, a "base tag" and the tag for the base tag+patches
14:08 Heffalump so then suspend would suspend just the patches covered by the second tag - first tag
14:11 pointfree[m] Yes, I would like to hide the effect of a version on the repo as a special case.
14:16 pointfree[m] still, at the moment "darcs tag" adds a marker and a "darcs stash" does not and would not... at least not as a record/commit.
14:26 bfrk interesting ideas
14:32 pointfree[m] afaik, at the moment, one shouldn't push and pull to and from repos with darcs stash having suspended some patches, but otherwise what Heffalump did is working from what I can tell.
14:33 Heffalump pointfree[m]: stash throws errors if you try to stash too many patches and they conflict too much (haven't tracked down the details yet)
14:33 Heffalump and the UI needs work, and the code really needs cleaning up
14:33 Heffalump I've been pretty busy at work for a while and haven't spent much time on darcs
14:39 pointfree[m] It would also need a way to store and list the subsets of patches, perhaps they could be stored as a list of weak hashes + their labels. After pushing to a label the weak hash would have to be updated.
14:41 pointfree[m] Heffalump: yeah, I'm rather swamped as well.
14:42 Heffalump pointfree[m]: I think two tags is enough, particularly if you just take the latest (chronological) tag with a given name if there's a conflict
14:55 pointfree[m] I had thought that the weak hash could be used to xor together any subset of patches not just the entire repo, and be used as a reference. Looks like it's just for comparing repos. Makes more sense. I guess I was confused.
15:00 Heffalump you could use a weak hash too, but I don't think it helps much. The biggest problem with any short identified in darcs is that it's a pain to find what it refers to without some kind of index
15:00 bfrk joined #darcs
15:21 favonia joined #darcs
15:29 Heffalump fabricandink: what version of tortoisemerge are you using/where can I get it?
15:35 bfrk joined #darcs
16:09 mizu_no_oto joined #darcs
16:57 bfrk joined #darcs
17:04 bfrk pointfree[m]: I think what you said about weak hash for a subset of patches (e.g. the 'stashed' ones) is a pretty nice idea and should work
17:32 Heffalump the tricky bit is getting from the weak hash to the actual set of patches
17:43 pointfree[m] I think an arithmetic sum of the hashes would be better than an xor of the hashes. It would require truncating or rounding off the number (or something). But it would still be more secure. For referencing subsets of patches smaller than the total number of patches the arithmetic sum would also be more useful, because you can do algebra with it.
17:53 Heffalump I don't see how - doesn't xor have all the same properties?
17:54 pem__ Heffalump, pointfree: I thought about this recently. There actually is a strong commutative hash
17:54 Heffalump commutative/associative - and it doesn't have a round-off/truncation problem so you get a better distribution of values
17:54 Heffalump in darcs you can't get strong hashes without commuting to the minimal context, IIRC
17:54 pem__ Just take the modular exponentiation of the sum of patch hashes
17:54 Heffalump that's why it's weak, primarily
17:54 Heffalump hashes of each individual patch
17:54 Heffalump xor might well make it weaker
17:55 pem__ do you mean patches change their hash when commuting?
17:55 Heffalump no, I mean you can only hash the metadata, not the content, to stop them changing their hash when commuting
17:55 Heffalump so it's a weak hash
17:55 pem__ I see.
17:55 Heffalump (dinner, biaw)
18:08 pointfree[m] yeah, forget what I said. It does have the same properties... and the subset sum problem is, afaik NP-complete.
18:52 mizu_no_oto joined #darcs
19:21 bfrk joined #darcs
19:23 bfrk Heffalump: of course, stupid me
19:47 bfrk left #darcs
20:18 lambdabot joined #darcs
20:33 lambdabot joined #darcs
23:07 favonia joined #darcs

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