Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Activity
    Fangrui Song
    @MaskRay
    cross-file-renaming has been supported for a long time.
    For clang-tidy, I am unsure that this must be integrated into a language server. Why doesn't a standalone service work?
    George Xanthakis
    @innerout
    I didn't know cross-file-renaming was supported that's great. Well about clang-tidy and ccls I feel like the errors reported by ccls diagnostics at the moment are only for basic things like syntax-errors (missing semi-colons etc.). With clang-tidy the diagnostics can spot bugs that you can easily miss. Also you don't need to leave your editor to run tools externally and search in thousand of lines reported by clang-tidy to spot the line you are interested.
    Dan Čermák
    @dcermak
    I just got a notification that a new version of ccls was released: 0.20190314.4, but hasn't that been released for over a year?
    and I've already pushed 0.20190314.5 to opensuse and Fedora ages ago, but that tag is gone now…
    What happened?
    Mark Poscablo
    @mark-poscablo
    Hi all, how do I add specific file formats to ccls' index whitelist (I'm working with C++) so that it could index those file formats too that aren't normally indexed/considered as traditional c++ files
    more specifically I'm trying to get ccls to index Qt's headers that are named "QApplication", "QWidget", etc, which of course doesn't have the ".h" suffix, but they include the actual ".h" header files and so act like an include name alias of some sort
    Fangrui Song
    @MaskRay
    @dcermak Force pushed so that 0.20190314.4 can be built with clang 9 and 0.20190823.6 can be built with clang 10 (there was an annoying handleDeclOccurence -> handleDeclOccurrence typo fix in the upstream which prevents 0.20190823.5 from being buildable with clang 10)
    Made use of this opportunity and moved some upstream adaptation as early as possible so that a wider range of commits can be buildable with clang 10
    Dan Čermák
    @dcermak
    @MaskRay Can you please, please, stop force pushing tags away
    this is really making my life as a downstream packager harder
    because now the tarball that I used half a year ago is no longer the same as what is up on github
    I'm begging you: please create new tags in a maintenance branch
    And don't get me wrong, I really appreciate that you try to keep ccls releases building with current clang versions, that's really great and helps me a lot
    but it would help me more if releases were truly immutable and I could just get the new tarball and rebuild things without having to wonder why the latest release is suddenly gone
    Fangrui Song
    @MaskRay
    I did not update existing tags. I just created new tags and force pushed to master.
    Dan Čermák
    @dcermak
    by force pushing to master you have modified the tags
    what I was meaning: please keep a tag assigned to one commit
    and don't change it ever again in the future
    it is okay if an old release no longer builds with a new compiler version
    that's when you create a new point release in a maintenance branch
    Fangrui Song
    @MaskRay
    I keep the commit ids of old tags. I force push to master and add new tags
    Dan Čermák
    @dcermak
    but 0.20190823.4 a day ago is no longer the same as it is today
    Fangrui Song
    @MaskRay
    I messed around the commits and probably made a mistake during the process... What was 0.20190823.4 before?
    I intended to create 0.20190314.4 but probably modified 0.20190823.4 unintentionally
    Fangrui Song
    @MaskRay
    Emmmmm. I don't think I touched 0.20190823.4 . I can't find anything related to it in my zsh history.
    Fangrui Song
    @MaskRay
    Confirmed. 0.20190823.4 did not change (condy0919/ccls@4534393 )
    Dan Čermák
    @dcermak
    @MaskRay apologies, I messed that one up, I meant 0.20190314.4
    that tag now points to very recent commits in contrast to what is around there
    and FTR 0.20190314.5 is no longer on master :-(
    Fangrui Song
    @MaskRay
    0.20190314.5 never exists I think :)
    Dan Čermák
    @dcermak
    a655ca96 - (tag: 0.20190823.5) indexer: disable warnings and skip processed function bodies (6 months ago) <Fangrui Song>
    argh, yeah mixed them up again
    I meant 0.20190823.5
    Fangrui Song
    @MaskRay
    0.20190823.5 has changed either. https://github.com/MaskRay/ccls/releases/tag/0.20190823.5 = https://github.com/condy0919/ccls/releases/tag/0.20190823.5
    David Manouchehri
    @Manouchehri
    How do I refresh semantic highlighting in VS Code without having to manually close and reopen the file? Is there a hotkey or something I'm missing?
    Christian Maniewski
    @chmanie

    Hey! I seem to be doing something stupid. ccls seems to be parsing all my .h files as C (as opposed to cpp, even if a corresponding .cpp files exists). I tried to add these lines to my .ccls file:

    %h -x
%h c++-header

    with no luck, I'm still getting unknown type name 'class'. Is there anything else I could try?

    The only way I can get around this currently is to rename the .h file to .hpp but that doesn't seem optimal to me (just for consistency with all the other libraries that I'm using and for which that doesn't seem to be a problem). For a little context, I'm using the auto-generated .ccls file provided by platformio
    Rob Blafford
    @graphcareful
    Hi not sure where to post this but i'll place it here. Just want to advise the contributors that i have found and fixed a potential issue. ccls has been crashing on me, upon further investigation I discovered that the --resource-dir parameter was incorrect. It was pointing to /usr/lib64/clang/9.0 instead of version 10.0.0. I have updated from fedora 31 to 32 (i.e. by way of the os update script)
    Rob Blafford
    @graphcareful
    @chmanie https://github.com/MaskRay/ccls/wiki/FAQ#full-indexing-takes-a-long-time--crushes-my-system
    omit your build/ directory or move it outside of your projects root dir
    if you just want ccls to parse some types of files, you can pass one or more regexs to initialBlacklist
    Christian Maniewski
    @chmanie
    @graphcareful thanks but I'm not sure how this relates to my problem. Sorry but I'm a bit confused here
    Patrick Ziegler
    @patrick96
    @graphcareful This happens when clang is updated without ccls being recompiled. Because ccls will query the current resource dir during compilation and will use that as the default.
    Christian Maniewski
    @chmanie
    I'm still not sure where my above issue originates from. I'm wondering if I could cut out ale (which I'm using as an LSP client) to see whether it's coming from there
    Christian Maniewski
    @chmanie
    OK, if I change %c -std=c++11 in the .ccls file it works. But that seems rather wrong. Shouldn't it detect it somehow?
    I thought that's what %h c++-header is supposed to do
    Christian Maniewski
    @chmanie
    It works now. Not sure what did it, but here's the whole config:
    clang

%h -x c++-header

%c -std=c11
%cpp -std=c++11
    Edward C
    @endinferno
    why my ccls cannot use std::of stream's function now? It can work before.