aboutsummaryrefslogtreecommitdiff
path: root/bash/bashrc.d
Commit message (Collapse)AuthorAgeFilesLines
* Suppress a couple of ShellCheck errorsTom Ryder2018-11-301-0/+1
|
* Remove redundant return short-circuitTom Ryder2018-11-301-1/+1
|
* Use ||/&& short-circuiting only for flow controlTom Ryder2018-11-301-19/+39
| | | | | If we're doing something besides return/exit, it should be a proper `if` condition so that the semantics are clearer.
* Stamp down three new ShellCheck 0.5.0 errorsTom Ryder2018-07-252-2/+2
| | | | All of these are false positives.
* Break sh noglob guard into two lines for clarityTom Ryder2018-06-292-2/+4
| | | | They're nicer to read this way.
* Quote all Bash completion targets literallyTom Ryder2017-11-081-45/+46
| | | | | | This helps ShellCheck understand them as literal strings for linting purposes, but it seems (and highlights) a little bit clearer this way anyway.
* Add "do", "then" keywords to Bash completionTom Ryder2017-11-061-0/+2
| | | | | | | | | | The Bash keywords "do" and "then" will be followed by another command. Adding them to this list means that pressing tab after "if foo ; then b" will complete for all command names beginning with "b". I was actually a little surprised that this worked, but there isn't really any reason to be; they're shell words just like everything else, not metasyntactic characters or anything like that.
* Break bash/bashrc.d/completion.bash inline listsTom Ryder2017-11-061-19/+66
| | | | | | | Break the sometimes long list of target keywords after the `complete` builtin and its options into newline-separated lists. This should make reading these lists a little easier, and also confuses Vim's syntax highlighting somewhat less.
* Escape backslashes in command prompt outputTom Ryder2017-07-051-4/+10
| | | | | | To stop them getting interpreted as \h, \w etc. None of these *should* be able to emit backslashes (it's an illegal branch name), but best to be thorough.
* Remove user@ prefix from promptTom Ryder2017-06-191-2/+2
|
* Allow setting a prompt mode for the shellTom Ryder2017-06-161-2/+2
|
* Revert "Use backslash to escape single char"Tom Ryder2017-02-171-2/+2
| | | | | | This reverts commit 245c2d6647e34fb432066360499ed02c5711271c. shellcheck worries about this.
* Use backslash to escape single charTom Ryder2017-02-121-2/+2
|
* Break a long lineTom Ryder2017-02-101-2/+2
|
* Break a long lineTom Ryder2017-02-101-1/+2
|
* Improve and correct some commentsTom Ryder2017-02-101-4/+4
|
* Start using Bash host completionTom Ryder2017-01-281-0/+4
| | | | But use ~/.hosts if it's there
* Extend "fallback" prompt a bitTom Ryder2017-01-271-1/+4
| | | | Include username/hostname if no SSH variables
* Show prompt prefix if a shell is exoticTom Ryder2017-01-251-0/+6
| | | | | | | That is, include e.g. "ksh:" as a prefix to the prompt if the user appears to have Bash or Zsh (or anything else) as their login shell. This is probably imperfect, but it's a start.
* Be more specific in ref name trimmingTom Ryder2017-01-251-1/+1
| | | | | e.g. for a namespaced branch "foo/bar/bar", don't strip the leading part off
* Remove SHLVL nested shell feature for nowTom Ryder2017-01-241-6/+0
| | | | | Causes a bit too much curliness in configuration; may need a more comprehensive approach.
* Add workaround to what looks like a bug to meTom Ryder2016-12-271-0/+5
|
* Tell ShellCheck to ignore an errorTom Ryder2016-12-271-0/+1
| | | | | ret actually is assigned, just not in any place ShellCheck could reasonably be expected to find
* Realised easy win avoiding subshellsTom Ryder2016-12-271-7/+4
| | | | And also the PROMPT_RETURN fudge in Bash/Zsh
* Use [!...] not [^...]Tom Ryder2016-12-221-1/+1
| | | | More portable
* Fix printf argument mismatch bugsTom Ryder2016-12-212-2/+2
| | | | What did we do before ShellCheck?
* Use clearer logic for prompt color settingTom Ryder2016-12-211-8/+7
|
* Request version with "ver" shortcut not promptTom Ryder2016-12-191-0/+4
|
* Attempt a much saner approach to managing SHLVLTom Ryder2016-12-191-3/+2
|
* Handle tmux sessions showing SHLVL properlyTom Ryder2016-11-301-3/+3
|
* Only add > starting from SHLVL=2Tom Ryder2016-11-301-1/+1
|
* Add > signs to show shell indentation levelTom Ryder2016-11-301-0/+7
| | | | | This needs extension to handle tmux sessions; it should be something like SHLVL - TMUX_SHLVL, which I'm not quite sure how to implement yet.
* Correct check of --is-work-tree testTom Ryder2016-10-151-1/+1
|
* Add % prefix to job spec completionTom Ryder2016-10-041-2/+2
|
* Stop `keep` writing empty files for undef varsTom Ryder2016-09-301-13/+12
|
* Completion for aliasesTom Ryder2016-09-231-1/+4
|
* Add some var quoting within param expansionTom Ryder2016-09-201-6/+6
|
* Block output/error from prompt tput(1)/git(1)Tom Ryder2016-09-201-102/+102
| | | | Should make things just a little bit faster
* Fix an issue (silencing Git prompt errors)Tom Ryder2016-09-081-1/+1
|
* Restore old code for describing Git prompt commitTom Ryder2016-09-071-1/+2
| | | | | | | | | | | | | | | I'm really confused. I could have sworn this was working correctly (i.e. the --always option spat out a short reference to the commit as a fallback), but I must have been wrong. Maybe I hadn't installed the appropriate file when I was testing it. At first I thought this was a bug introduced in Git 2.10, but the short-circuit disregarding --always if --exact-match or --candidates=0 is set seems to have been in the `describe` builtin for ages. I must have just tested poorly; I can't see any other explanation in the source. No matter; this isn't much slower.
* Complete kill builtin with jobspecs and user PIDsTom Ryder2016-09-051-1/+1
|
* Flesh out builtin completions a littleTom Ryder2016-09-051-7/+8
|
* Test number of colors properly for promptTom Ryder2016-09-031-31/+26
|
* Correct completions for declare, typeset etcTom Ryder2016-09-011-4/+1
| | | | All of them can accept function names
* Restore git symbolic-ref as first attemptTom Ryder2016-08-311-2/+4
| | | | | Otherwise if two branches point to the same commit, you can't be sure which one will show up.
* Efficiency tweaks for Subversion promptTom Ryder2016-08-311-17/+11
|
* Stop loop when we have needed svn info varsTom Ryder2016-08-311-3/+2
|
* Remove unneeded fallback for $nameTom Ryder2016-08-311-2/+1
| | | | We've already bailed out by this point if we couldn't get one
* Move git index update higher upTom Ryder2016-08-311-3/+3
|
* Use simpler method to find branch nameTom Ryder2016-08-311-11/+9
| | | | | Only one call to git(1), too; seems to work at least as far back as Git v1.5.6.5