Just my two cents, but InstaREPL in its basic form was just unusable for any serious development beyond terse one-off oneliner expressions. It made people lose focus on what is important (building expressions) to draw attention to unhelpful error messages. Which was 9/10 times due to expressions not being completely written yet. It was just a bad idea and any delay timer wouldn't help it either since everyone types at a different pace, expressions have varying length and I could go on and on about how it sucked. The current, I would say close cousin, is much much better. It requires intelligence (a human) to determine when done with formulating the expression and no algorithm currently known to man can accurately predict that based on incomplete input. Even if it would *seem* done, it doesn't have to be: one can cut/copy/paste expressions around in a document and only then be done so while a expression might be evaluable, say when parens closes and are matched (balanced expression) doesn't make it so that it is 'done'.
Second point, autocomplete works here albeit lagging now and then. But clojure core functions are displayed in the drop down, so are those functions from external libraries and when I need/want to know what arguments it accepts and in what order, I use `ctrl-d` to get the docstring up. Much much faster, easier and better than any 'press F1, wait to open up web browser, navigating to vendor page' or even `alt-tab` for that matter. So in the end, I fully support the idea to remove InstaREPL from LT because it promoted bad practice and I feel the core of Clojure is to have developers focus on the important stuff, and less on the fluff/distractions and cognitive workload e.g. OOP/imperative programming tends to burden developers with. The complete opposite of Visual Studio and .NET architecture, to name one place where you need to remember hundreds of seemingly same classes and methods that have tiny differences nobody really cares about but become important to distinguish. My only grief currently, is the terrible windows experience I am having with LT and I really need it to run there due to data security and privacy policy. On Saturday, December 31, 2016 at 4:41:33 PM UTC+1, afro54 wrote: > > Hi, > Please excuse any sense of whining that you get from this - I just want to > know if I can configure LT to be as useful as it once was. > > 1) I badly miss the instarepl and, despite my efforts, I haven't been > able to replicate the functionality. > Instarepl is available in Nightcode but, the app isn't quite there in > terms of multi-tab editing and a few other (minor) features. Sometimes, > I'm using LT to launch NC, which points to my frustration. > 2) I badly miss code-completion. Why on earth do I fail to even get this > feature? It means that, the more I try to explore Clojure software, the > more I have to try and remember. > > Essentially, all the good features that caused LT to raise so much more > money that Chris Granger wanted, have been disabled. Is that the case, or > can things be made better? > > I can connect to nrepls that I launch - this helps very little. > > Auto-complete works on my .behavior files but, not my .clj files - what is > that all about - I really need to fix that. > I require dependencies but, receive no documentation support in using > them. > > I ask because I am watching this tutorial using Cursive and IntelliJ, and > it makes me want to cry, that setup seems so productive to use. > > Is there any definitive method to get Light Table to be impressive once > again? Any configurations that can restore my pride in the LT environment? > > Even more generally, if people are dumping LT, despite being fans of LT, > what alternative should I consider? > > Thanks, all. > -- You received this message because you are subscribed to the Google Groups "Light Table Discussion" group. To unsubscribe from this group and stop receiving emails from it, send an email to light-table-discussion+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
