Sender | Message | Time |
---|---|---|
22 Dec 2020 | ||
mesajor | [rsteube(gitter)] yes i had that, but had some problems with that - getting pretty close now though: | 22:58:10 |
mesajor | [rsteube(gitter)] YAY | 23:32:30 |
26 Dec 2020 | ||
mesajor | [Ian(telegram)] Did one of the commits today break @zzamboni’s alias module? Seems if I have any aliases defined it’ll fail to start. | 00:58:12 |
mesajor | [Ian(telegram)]
| 00:59:06 |
mesajor | [Ian(telegram)] bash-3.2$ elvish panic: runtime error: index out of range [4] with length 4 goroutine 300 [running]: github.com/elves/elvish/pkg/eval.fnOp.exec(0x4, 0x152f580, 0xc0000d2460, 0xc0002eb7a0, 0x1, 0x1) github.com/elves/elvish/pkg/eval/builtin_special.go:188 +0x2d7 github.com/elves/elvish/pkg/eval.(*formOp).exec(0xc0000de210, 0xc0002eb7a0, 0x0, 0x0) github.com/elves/elvish/pkg/eval/compile_effect.go:341 +0xdf9 github.com/elves/elvish/pkg/eval.(*pipelineOp).exec.func1(0x152acc0, 0xc0000de210, 0xc0002eb7a0, 0xc000188288, 0xc00061c480, 0x17c3200) github.com/elves/elvish/pkg/eval/compile_effect.go:150 +0x3f created by github.com/elves/elvish/pkg/eval.(*pipelineOp).exec github.com/elves/elvish/pkg/eval/compile_effect.go:149 +0x225 (edited) | 00:59:19 |
mesajor | [Ian(telegram)] If I remove any aliases from ~/.elvish/aliases/ or just comment out the use github.com/zzamboni/elvish-modules/alias line in ~/.elvish/rc.elv everything is fine. | 01:00:09 |
mesajor | [krader1961(gitter)] Possibly. The most likely being https://github.com/elves/elvish/commit/6419f4524a1a5217e5d29fdc9ff0b6d4ca13b1a7. I no longer use @zzamboni’s alias module so I can’t readily confirm. Try building without that commit. Regardless, open an issue. | 01:00:11 |
mesajor | [Ian(telegram)] Well, is it a bug in elvish or a breaking change that the module needs to adapt to? | 01:00:55 |
mesajor | [krader1961(gitter)] It could be both but it’s a bug because it’s causing a panic. | 01:01:42 |
mesajor | [Ian(telegram)] Fair enough | 01:01:52 |
mesajor | [Ian(telegram)] https://github.com/elves/elvish/issues/1202 | 01:07:41 |
mesajor | [Ian(telegram)] I wind up using aliases to define simple shortcuts, ll for ls -lh or things like that. Is there a better way to do that? | 01:09:39 |
mesajor | [krader1961(gitter)] I don’t know about better but I decided to eliminate the middleman. I put my functions I want to be easily usable in a couple of modules that I import in my ~/.elvish/rc.elv. Then I alias them into my interactive namespace:
| 01:11:39 |
mesajor | [krader1961(gitter)] That is essentially what @zzamboni’s alias module does. | 01:12:19 |
mesajor | [krader1961(gitter)] One minute while I confirm that my approach works in light of the changes merged today… | 01:13:11 |
mesajor | [Ian(telegram)] Yeah, I was trying to avoid a massive rc.elv, but it doesn’t really matter either way I suppose. | 01:13:13 |
mesajor | [Ian(telegram)] Ah, for me it is failing when I just use github.com/zzamboni/elvish-modules/alias with any aliases defined on disk. The export line is fine (well, apart from the failure to get to that point…) | 01:13:57 |
mesajor | [krader1961(gitter)] Gah! My approach also fails with a panic. I see @xiaq has made some other changes today not tied to an open issue. Reverting to HEAD~3 doesn’t panic but does result in this error: "Exception: compilation error: cannot find variable $edit:insert:binding”. | 01:23:23 |
mesajor | [krader1961(gitter)] If I revert the next to commits (including the one I mentioned earlier) then I get a working interactive shell. | 01:24:34 |
mesajor | [Ian(telegram)] Interesting. Oh well, not a big deal for the moment. | 01:30:06 |
mesajor | [krader1961(gitter)] I’d say it’s a pretty big deal because the five most recent commits result in an unusable interactive shell. | 01:30:40 |
mesajor | [Ian(telegram)] Well, if you’re using a - feature ;) | 01:30:58 |
mesajor | [Ian(telegram)] Anyways, it isn’t something that needs to be fixed now, but is something that should be fixed soon. | 01:32:55 |
mesajor | [krader1961(gitter)] About that… I would argue that whether the $-exports- interactive namespace var is technically experimental it is the only mechanism for defining interactive functions that doesn’t require putting all those function definitions in you ~/.elvish/rc.elv script. Either that var needs to be renamed so it is no longer technically experimental or an equivalent, non experimental, mechanism for doing the same thing is needed. | 01:35:43 |
mesajor | [kolbycrouch(gitter)] Hi all. Anyone notice that $args no longer works on latest master? elvish -c 'put $args' 1 2 returns [] . On my old tree from a few months ago it works as expected. | 09:12:03 |
mesajor | [zzamboni(telegram)] Hi everyone - back into Telegram after a loooong while. I have posted my latest findings from the current panic at https://github.com/elves/elvish/issues/1202 | 12:05:09 |
mesajor | [zzamboni(telegram)] I disagree, at least I rely heavily on aliases. I could redefine all of them as functions, but… (re @Ian: Anyways, it isn’t something that needs to be fixed now, but is something that should be fixed soon.) | 12:06:38 |
mesajor | [Ian(telegram)] Ah, I look at my aliases as “nice to have”, mostly because I usually don’t have aliases defined on a lot of stuff at work. | 12:25:07 |
mesajor | [Ian(telegram)] It would be nice to have -exports be a non-experimental thing, or have some sort of source -like ability. | 12:25:49 |
mesajor | [xiaqqaix(telegram)] seems i broke -source | 14:42:32 |