Bug 69200 - [UX] Pervasive "find" box/field for lists/trees
Summary: [UX] Pervasive "find" box/field for lists/trees
Status: NEW
Alias: None
Product: Platform
Classification: Eclipse Project
Component: UI (show other bugs)
Version: 3.0   Edit
Hardware: All All
: P3 enhancement with 6 votes (vote)
Target Milestone: ---   Edit
Assignee: Platform UI Triaged CLA
QA Contact:
URL:
Whiteboard:
Keywords:
: 195579 201319 224208 224493 312043 321586 322788 (view as bug list)
Depends on:
Blocks: 304440
  Show dependency tree
 
Reported: 2004-07-02 07:29 EDT by Timo A. Hummel CLA
Modified: 2019-02-11 09:21 EST (History)
22 users (show)

See Also:


Attachments
Fake-example of my feature request (31.88 KB, image/jpeg)
2004-07-02 07:30 EDT, Timo A. Hummel CLA
no flags Details

Note You need to log in before you can comment on or make changes to this bug.
Description Timo A. Hummel CLA 2004-07-02 07:29:57 EDT
In large projects, my directory and file trees are becoming very large. It would
be nice to have a feature like Mozilla's search feature where you type in the
web site and the browser finds the next string.

Similar, it would be nice if eclipse would provide such a functionality for the
navigator. Just click in the tree, and start to type - the navigator should find
the next ocurrence of the string entered (matching also substrings, and not from
the beginning as it works now).

See the attached fake image.
Comment 1 Timo A. Hummel CLA 2004-07-02 07:30:32 EDT
Created attachment 12940 [details]
Fake-example of my feature request
Comment 2 Nick Edgar CLA 2004-07-05 12:07:21 EDT
In Firefox or Mozilla, how do you advance to the next match?  Do you have to
wait for it to time out and type the substring again, or is there a key to advance?

For navigating around large projects, I recommend using Ctrl+Shift+T (Open Type)
and Ctrl+Shift+R (Open Resource), optionally with Link with Editor on in the
Navigator and/or Package Explorer.  These prompters have type-ahead filtering
(prefix matching by default).  If you want to match a substring use, e.g.
*substring*.


Comment 3 Timo A. Hummel CLA 2004-07-05 12:14:34 EDT
Mozilla and Firebird are advanding to the next match by hitting F3.

I did not knew about "Open Resource" so far - it looks very promising. Could it
made possible that it matches substrings by default (or by configuration?)

Comment 4 Kevin McGuire CLA 2007-08-28 14:00:19 EDT
Came across this bug while searching for another.  I've recently come to the same conclusion that we should make this support pervasive since the lists/trees are getting huge.  It'd be cool if you got it for free for any list/tree viewer, but at the least it should be easy to plug.
Comment 5 Mike Wilson CLA 2007-09-18 09:02:55 EDT
Assigning to Kevin: I wouldn't this in the 3.x stream, but it feels like this notion (i.e. that trees and lists must always be prepared to show a subset of their contents) needs to be part of our model for the workbench in 4.0.


Comment 6 Kevin McGuire CLA 2007-09-24 15:00:33 EDT
Agree, good 4.0 item.  I'm marking it as such so we can track it.
Comment 7 Martin Aeschlimann CLA 2007-09-28 04:21:43 EDT
The interesting question for this feature is: Can it find elements inside nodes that are not yet expanded. I assume you want it to do this. Problem is that this is an expensive operation. For example in the package explorer this would mean to open lots of Java files and JARs.

Finding an element by name in a large model is something you better do with an index. We have 'Open Resource' and 'Open Type' which contain a lot of infrastructure to provide a good user experience (search in the background, history for immediate results)
Use 'Navigate > Goto Resource', 'Navigate > Goto Type' to reveal an element in the package explorer/navigator
Comment 8 Gregor Rosenauer CLA 2008-02-11 14:58:35 EST
(In reply to comment #7)
> The interesting question for this feature is: Can it find elements inside nodes
> that are not yet expanded.
Would it help if the search only accounted for previously expanded (i.e. cached) items? This would avoid the additional cost of a "deep" lookup and meet most user's expectations. Also, I would not expect it to search *inside* JARs, I only want a quick way to get to the artefacts in my project.

> Finding an element by name in a large model is something you better do with an
> index. We have 'Open Resource' and 'Open Type'
Agreed, but as users get used,-) to quicksearch in dialogs, they'll expect this functionality in other areas too, where it makes sense.

Curious about 4.0... is there anything on the wiki yet?
Comment 9 Kevin McGuire CLA 2008-02-15 12:57:46 EST
(In reply to comment #8)
> Would it help if the search only accounted for previously expanded (i.e.
> cached) items? This would avoid the additional cost of a "deep" lookup and meet
> most user's expectations. Also, I would not expect it to search *inside* JARs,
> I only want a quick way to get to the artefacts in my project.

The key in my view is to automate the case where I know the thing is there in the tree, but visually scanning it is tiresome. So that matches your suggestion.  

On the other hand, it sure would be tedious if it failed because I hadn't expanded a /src folder yet. More so, it'd be confusing since a successful hit depends on whether I've seen the element already (since I started Eclipse today? ever in that workspace?), but my memory is only so good, so now I can't tell if a failure is because I've not yet seen it or it doesn't exist.

So I think it needs to be driven off an indexer.  But we may bound it to resources that exist in the workspace via a loaded project vs. resources which are referenced (e.g. within a JDK jar or an installed plugin).

> Curious about 4.0... is there anything on the wiki yet?

Not as yet.

Comment 10 Susan McCourt CLA 2009-11-30 16:34:58 EST
Retitling bug (again).
We tend to use "Filtered Tree" for these cases today, but I didn't mean to imply that a pervasive quick find would actually filter the list.
Comment 11 Susan McCourt CLA 2009-11-30 16:36:21 EST
*** Bug 201319 has been marked as a duplicate of this bug. ***
Comment 12 Susan McCourt CLA 2009-12-16 14:13:03 EST
This is a bit different than the original proposal (a search bar on each view), but I thought I should mention it here.  There is talk of having a search bar in the trim of an e4-based workbench.

One idea is that it could do this style of "find" for the active part.
Another idea from Bogdan was that it could be a good home for "quick access" style finding (Ctrl+3).  So that you could always type the name of any command/dialog/part you were looking for and quickly find it.

In the first case you are looking for "something to work with."  
In the second case you are looking for "something I want to do/open".
Comment 13 Dani Megert CLA 2010-03-26 09:14:18 EDT
*** Bug 224208 has been marked as a duplicate of this bug. ***
Comment 14 Dani Megert CLA 2010-05-07 09:17:44 EDT
*** Bug 195579 has been marked as a duplicate of this bug. ***
Comment 15 Dani Megert CLA 2010-05-07 09:20:03 EDT
*** Bug 312043 has been marked as a duplicate of this bug. ***
Comment 16 Dani Megert CLA 2010-08-16 12:29:00 EDT
*** Bug 322788 has been marked as a duplicate of this bug. ***
Comment 17 Markus Keller CLA 2010-08-18 07:18:56 EDT
The index-based search from comment 7 and comment 9 requires each view to provide the infrastructure to efficiently search in an index of their model elements. Hence, this functionality cannot be added pervasively -- only view-by-view.

I agree with Kevin that a search that considers all "previously expanded items" (comment 8) would be unsatisfactory.

But what about a search that only considers visible items? That would not solve all problems, but it would help in many cases, would be easy to explain, and can be implemented for all views using only information from the label provider (so it matches exactly what the user sees). Ctrl+F on GTK+ already does something like this -- the only thing it misses is wildcard matching (except for a free * at the end).

(In reply to comment #0)
> [..] Just click in the tree, and start to type [..]

I wouldn't change the platform behavior when the user just starts typing. But we could support Ctrl+F to start with an empty search pattern and the * key to start with "*" (so the user can easily search for substrings).
Comment 18 Markus Keller CLA 2011-02-24 09:54:32 EST
*** Bug 321586 has been marked as a duplicate of this bug. ***
Comment 19 Dani Megert CLA 2019-02-09 12:09:00 EST
*** Bug 224493 has been marked as a duplicate of this bug. ***