I love Visual Studio Code, but its Intellisense auto-complete drives me crazy and I make more typos with it than it helps. So I feel I must be using it wrong.
The problem is very hard to explain, so I have a screenshot below:
I typed in thi. I would hope/expect/want that autocomplete would only look for anything to autocomplete that contains thi in consecutive order. But it does not. Instead it looks for anything with the letters t, h, and i in them. They needn't be next to each other, nor does the thing of interest even need to start with t.
I would like to "tame" auto-complete to only find consecutive letters. Is there a way to do this? (I use Python, Javascript, and SQL for most of my work, but I'm hoping the config is cross-language.)
Ideally, I would like the auto-complete to (a) require all letters be consecutive, (b) not require the thing of interest to start with those letters, and (c) ignore upper/lower case. But by far the most important issue to me is resolving (a).
The issue you are having is a feature added in November 2017 update. There is currently no way to turn off fuzzy autocomplete, but the issue is currently open on GitHub and the setting to change this behavior will be added.
In the meantime you can tweak your autocomplete with these settings:
"editor.wordBasedSuggestions": false,
"javascript.nameSuggestions": false,
"editor.snippetSuggestions": "bottom" / inline, none
"editor.suggestSelection: "recentlyUsedByPrefix"
From official docs:
When using the last option, recentlyUsedByPrefix, VS Code remembers which item was selected for a specific prefix (partial text). For example, if you typed co and then selected console, the next time you typed co, the suggestion console would be pre-selected.
This lets you quickly map various prefixes to different suggestions,
for example:
co -> console and con -> const.
And maybe:
"editor.quickSuggestions": {
"other": false,
"comments": false,
"strings": false
},
It doesn't solve your problem completely, but it's the closest you can get without digging through VS Code source and creating an extension.
Related
I'm in PhpStorm and I need to select some text, press some shortcut and have that text wrapped in a function call (that I would have defined somewhere in the settings beforehand).
For example:
"Hello World" would become input("Hello World").
$_GET["foo"] would become input($_GET["foo"]).
I don't know if this is even possible, but it could help me save so much time if so.
Applying regex to solve this problem is unfortunately not possible. Manually selecting what I need to wrap isn't an issue.
Will the function name be the same every time or different?
In any case: it can be done this way:
Make a Live Template of "surround template" type with the following content:
$FUN$($SELECTION$)$END$
The $SELECTION$ variable here tells that it's a surround template.
Apply correct Context (where this template can be used)
Give it an abbreviation (name used to locate & invoke it) and brief description.
Here is mine:
NOTE: replace $FUN$ by a fixed function name if the function will always be the same. You can have additional templates with different abbreviations (that will have different hardcoded function names).
To use it:
Make a selection and invoke Code | Surround with... action where you select the right template. On Windows keymap it's Ctrl + Alt + T
In action (NOTE: it's without hardcoding the function name hence me typing the myFunc part):
(HINT: you can select the desired entry in a few keypresses if the name is unique -- just start typing the name in the popup -- the standard Speed Search work here)
P.S. Code | Surround with... can have other (possibly irrelevant for you in this case) entries. To list Live Templates only, use the shortcut for Surround with Live Template... action (Ctrl + Alt + J here on my Windows keymap). You can check the shortcut or change it in the Settings/Preferences | Keymap:
This way the popup menu will be a bit shorter:
Less keypresses:
You may be able to use Macros functionality to record the invoking the popup and selecting the right entry. You can then assign a custom shortcut to that Macros: select the text, hit the shortcut and it will playback the recorded sequence.
Sadly I cannot 100% guarantee that Macros will always work nicely (sometimes/on some setups it can "eat" keypresses).
P.S. It would be much easier if the IDE would support assigning keyboard shortcuts to specific Live Templates .. but it's in the backlog and no ideas on when this might be implemented. Anyway: https://youtrack.jetbrains.com/issue/IDEA-67811 -- watch this ticket (star/vote/comment) to get notified on any progress.
P.S. You can also try Postfix completion. It's good for writing the code and not really suitable for your case (editing small parts of it), but who knows. You will have to make a custom postfix for this -- should not be an issue though.
https://www.jetbrains.com/help/phpstorm/2021.3/auto-completing-code.html#postfix_completion
Does anyone have any luck adapting Draft.js for Reagent? There's pretty heavy editing issues if Draft.js is imported right away via reagent/adapt-react-class. Cursor jumps, disappearing symbols when you're typing, onChange calls with incorrect EditorState, you name it.
People are reporting problems like this in clojurians/reagent Slack channel, but it seems there's no solution so far.
Any help would be greatly appreciated.
Okay, thanks to tonsky, I got the answer. Reagent/Rum are using deferred rendering with requestAnimationFrame, but Draft.Editor should be re-rendered immediately when editorState is set.
All we need is to call forceUpdate for editor parent component whenever editor onChange is invoked:
:editorState #editor-state-atom
:onChange (fn [new-state]
(reset! editor-state-atom new-state)
(.forceUpdate #wrapper-state))
Code example is for Reagent, solution for Rum is identical
Just an idea (more of a comment, but I can't provide comments yet), because -- if I remember correctly -- the :content-editable attribute is treated in a special way within Reagent:
Since the problems seem to occur when the Draftjs editor is called from Reagent, would it not help to convert the Reagent-component to a React-component (using reagent/reactify-component), and then use this "reactified" component as a react-component within Reagent, using reagent/create-element? I would assume that Reagent then refrains from meddling with the Draftjs editor.
In VS 2015, only when in Razor (.cshtml) files, roughly half of the time the autocomplete/suggestion list/intellisense doesn't work correctly (sorry, not sure the actual term... when you type an object and hit . and the list of properties and methods shows to select from)
The behavior is that when I hit ., the list popups up for a fraction of a second and then closes. It happens so fast I try to do a quick Backspace, ., Backspace, . cycle a few times to at least see the name I need, but I usually cant' get it and end up having to find the exact name elsewhere from code. Extremely irritating...
It happens sporadically with no real pattern I can find. Here's patterns that I've ruled out:
The file that's open doesn't seem to matter.
Whether or not I close/reopen the file doesn't seem to matter
Whether I navigate to another file and back doesn't seem to matter
It will work/not work multiple times on and off throughout the same file
It doesn't seem to be relevant to any particular object/property/method
I've checked all my options (there doesn't seem to be Text Editing options for Razor?), have tried clearing caches, the reloading solution/projects, restarting VS, all of which seem to still provide no pattern.
Has anyone come across this and have any ideas of where else I can look to fix it?
Example
Here's an extremely simple example... new project, very little code/files, very simple view. Where the Model. stops, I should have the usual base methods, and an 'Items' collection. It pops up for a split second then disappears... no lambdas/complex view parsing involved (this is reproducible as well):
Update: Patterns
Things I've noticed:
If I'm entering a #model ns.ns.ns.type, it rarely happens toward the "base" end of the namespaces. It's as I get further towards the type that it happens. This one is intermittent.
In some cases, it works perfectly fine, every single time. For example, I often use DevExpress tools, and have never seen the behavior on any of their extensions (so, #Html.DevExpress(). (and other similar, not necessarily DevEx models) will never cause a problem)
It happens almost all the time when I'm accessing my #Model (which is where I most want it!). I've found some cases where this is reproducible every time (see above example), but it's about 90%+
Occasionally, as I work through the object tree, one will fail while the next works (ex: #models ProjName.Web.App.Subscriptions.Models.AccountCreateVM... it might fail on Subscriptions but work fine on Models)
Occasionally, beginning to type the name within autocomplete kicks it back into gear and it starts working again. In the above example, starting to type Acc for AccountCreateVM causes it to start working again.
I haven't found the root cause, but in all cases, CTRL+SPACE works. This isn't the best, but light years better than nothing at all.
(this shortcut is not one I've used in the past, so this is likely standard behavior, but...) If you're at the dot Model. and autocomplete list disappears, CTRL+SPACE consistently brings it back up, and when it does come back, it stays! If there's only one possible autocomplete member, it'll auto-fill the member for you upon CTRL+SPACE
This happens for me all throughout VS2015 during lambda statements.
It happens when editing code "mid-document", as in, if there is anything besides a ) or } following where I'm typing. VS appears to be struggling to tell where the current statement ends & the next statement begins.
The following code will consistently fail to trigger Intellisense at the period, even when explicitly invoked.
var subset = initialSet.Where(x => x.
var result = new Whatever();
In Razor, it is very common to be editing code between existing text and using lambda statements:
<strong>#Html.DisplayFor(m => m.</strong>
This is probably why you only experience this in Razor.
The way I work around this bug is just to write the ) to close the method.
var subset = initialSet.Where(x => x.)
var result = new Whatever();
<strong>#Html.DisplayFor(m => m.)</strong>
Intellisense can then be triggered on the period.
If you're using a method that requires a minimum of more than just the lambda (like RadioButtonFor), you'll also need to put in a comma for each of the extra parameters.
<strong>#Html.RadioButtonFor(m => m.,)</strong>
If Intellisense is appearing, but immediately disappearing again, the best solution I've found so far is to just type a few letters of any known member, then using Ctrl-Left to skip back to the period, and trigger Intellisense again (Ctrl-Space or backspace-retype). This usually gets it to appear and stay. You'll have to delete the characters you typed afterwards, which can be frustrating.
Just make sure the ) does not touch the text you are editing, and the popup will stay up.
Instead of...
#Html.Partial("ManageGrid", Model.)
Use...
#Html.Partial("ManageGrid", Model. )
The intellisense seems to get confused by touching close parenthesis. Not ideal, but this was the only way I could get it to work for me consistently.
In my specific case, i was able to solve the problem by installing the latest version Microsoft ASP.NET and Web Tools.
https://marketplace.visualstudio.com/items?itemName=JacquesEloff.MicrosoftASPNETandWebTools-9689
Once I installed it, the problem was gone. It is likely that this update fixed something that could be fixed with an older version, but either way I'm happy.
(I found this in Visual Studio under Tools->Extensions and Updates...->Updates->Visual Studio Gallery)
I had the same error and I fixed it by deleting all the files of the component model cache.
This is the path:
Users\YourName\AppData\Local\Microsoft\VisualStudio\14.0\ComponentModelCache
Hope that helps
I use ctrl+j as a temporary solution when I know the content.
Or keep writing without right parenthesis can use the completion:
#Html.LabelFor(m => m.Name
Whenever this annoying thing happens to me, I just put an extra dot and then it works. I have to put the extra dot every time. For example, if I write this and intellisence flashes and disappears:
#Html.TextBoxFor(m => m.
then I just do this:
#Html.TextBoxFor(m => m..
And intellisense will now show after first dot. I have made this a habit until MS has a fix for it.
Instead of...
#Html.Partial("ManageGrid", Model.)
Use...
#Html.Partial("ManageGrid", Model.
In PhpStorm 9.0.2 (I do not know about the newer versions, but I am afraid it will be the same) when you search for something and get the results in the Find tab, you can mark certain results as irrelevat by selecting them and pressing the del key and this will strike them out.
When you search for something very common, which can have 5000+ occurences, even by striking some results out, the Find tab becomes very bloated.
Is there a way to completely remove a result from the Find tab, instead of just striking it out?
When using Find/Replace in Path functionality and you know that you will get a lot of results it's better to limit your search scope by either specific path, specific file extensions or some custom scope (using Scopes functionality).
If such "limiting" cannot not help (e.g. you really need to go trough each occurrence) then use Alt + Delete (the Remove action).
Please note that this action is most likely available since PhpStorm v10 only (it was added only recently).
You can check if you have it and what the shortcut is via content menu:
You can use this keymap shortcut : ALT + SUPPR (this remove the result out of the find tab)
instead of
keymap shortcut : SUPPR (this JUST exclude the result of the find tab with underline this not remove)
EDIT :
This is the exact response for your issue. Look at this screenshoot which say all :)
I am writing a Google Chrome Extension and trying to make the transition to coding in HTML, CSS and Javascript from the kind of coding I did 40 years ago. It's a big learning curve but a little bit of console logging would help a lot. (Who remembers the green phosphor displays on the old iron-core memory Sperry Univacs? Now there was a console that could tell you what was going on - mount a tape, change a diskpack or put more paper in the printer.) But I digress.
Just a few days ago, I was able to send messages to the Google Chrome DOM Inspector console log from my content script or background page like this:
console.log="RECEIVED REQUEST FROM CONTENT SCRIPT PEEK.JS";
Later if I inspected the page where that code was injected, I would see that exact same message on the console. If that line executed four times, I would see four instances of that messages in sequence on the console. It was simple but beautiful.
Now, nothing appears on the console unless I type in "console.log" and in this case I see only the last instance of the message. What's the problem here? Did I toggle something off in the debugger? Am I writing the code wrong? Should it be: "console.log("Message Here");" - (this always returns an error) ?
I have spent literally hours trying to find the answer to this question and I can only surmise that the question is so simple that even a well-educated child could answer it. The alternative is to spend more hours trying to find my bugs thru implication or from alert messages. Better to spend my hours learning how to effectively use the inspector for debugging.
For example, yesterday after restructuring what was once working code so that it would pass an array to background from content instead of background requesting each item successively from content, the extension broke, as might be expected. The console displays one error message: "Error during tabs executeScript: Unknown error." with a link that says "chrome/ExtensionProcessBindings:95" on the far right of that same line. Obviously the error message is almost useless. Clicking on the link brings me to the resources panel which displays a blank frame to the right with the words "background.html" This appears to be useless as well.
How is this information at all helpful? It took me another six hours to figure out that I had moved a variable definition to another place in the code so that it was not defined at the time the executeScript line was executed. Had the error message said that a parameter was undefined in executeScript, I could have found it in about ten minutes.
So back to the original question - I WANT A SIMPLE ANSWER - don't tell me to write a function. I saw the console displaying what I wanted two days ago without writing any functions.
Here is your SIMPLE ANSWER: Instead of
console.log="RECEIVED REQUEST FROM CONTENT SCRIPT PEEK.JS";
You should use
console.log("RECEIVED REQUEST FROM CONTENT SCRIPT PEEK.JS");
For console.log():
https://developer.mozilla.org/en-US/docs/Web/API/console.log
I would recommend searching the Mozilla Developer Network for documentation on JavaScript and many browser technologies.
My general suggestion to help with mistyped or out of scoped variables is to use strict javascript:
https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Functions_and_function_scope/Strict_mode
This makes certain JavaScript idioms into errors which are detected earlier, and which give more helpful diagnostics. (The quick version is to put the single line "use strict"; at the top of your js files, or if that causes to many problems you can put it as the first line inside any function to do function-by-function migration.)
The reason why console.log = blah would show blah on the console in some situations is that in JS, <left side> = <right side expression> is an expression which evaluates to the same value as just <right side expression>, but it has the side affect of also assigning to <left side>. (This is similar to C assignment expressions.)
Then, in the context you were in, the console will display the result of any expression evaluation. This is different from explicitly calling console.log() which will display the arguments as the intended side effect of that function.
So in that case you were overwriting the standard console.log function with a string, and the entire expression evaluates to that string value, and the console's builtin behavior (not the console.log() function!) would display the result of your evaluation.
In JS most things are mutable, and you can overwrite all kinds of fields and variables, so be aware of this potential gotcha!
Also, a general recommendation for seeking help on the web:
For sites like StackOverflow is to limit your post to a very specific single question. This is because people with similar problems will come here based on web searches or other links, and they need to figure out if this page helps them with their particular problem quickly. If they have to read through paragraphs with multiple questions or anecdotes, this really slows them down. I would guess that's why this question has been downvoted.
If you have multiple questions, make separate posts. StackOverflow isn't a good site for anecdotes about your experience in the past. FWIW- I personally found your background as a programmer on 70's era machines interesting, but many readers may get lost in the details. Those kinds of anecdotes could make a good blog post.