Mercurial > vim
view runtime/doc/quickfix.txt @ 1976:27060fe297d0 v7.2.273
updated for version 7.2-273
| author | vimboss |
|---|---|
| date | Tue, 03 Nov 2009 13:26:55 +0000 |
| parents | 5232b9862f23 |
| children | de5a43c5eedc |
line wrap: on
line source
*quickfix.txt* For Vim version 7.2. Last change: 2008 Mar 14 VIM REFERENCE MANUAL by Bram Moolenaar This subject is introduced in section |30.1| of the user manual. 1. Using QuickFix commands |quickfix| 2. The error window |quickfix-window| 3. Using more than one list of errors |quickfix-error-lists| 4. Using :make |:make_makeprg| 5. Using :grep |grep| 6. Selecting a compiler |compiler-select| 7. The error format |error-file-format| 8. The directory stack |quickfix-directory-stack| 9. Specific error file formats |errorformats| {Vi does not have any of these commands} The quickfix commands are not available when the |+quickfix| feature was disabled at compile time. ============================================================================= 1. Using QuickFix commands *quickfix* *Quickfix* *E42* Vim has a special mode to speedup the edit-compile-edit cycle. This is inspired by the quickfix option of the Manx's Aztec C compiler on the Amiga. The idea is to save the error messages from the compiler in a file and use Vim to jump to the errors one by one. You can examine each problem and fix it, without having to remember all the error messages. In Vim the quickfix commands are used more generally to find a list of positions in files. For example, |:vimgrep| finds pattern matches. You can use the positions in a script with the |getqflist()| function. Thus you can do a lot more than the edit/compile/fix cycle! If you are using Manx's Aztec C compiler on the Amiga look here for how to use it with Vim: |quickfix-manx|. If you are using another compiler you should save the error messages in a file and start Vim with "vim -q filename". An easy way to do this is with the |:make| command (see below). The 'errorformat' option should be set to match the error messages from your compiler (see |errorformat| below). *location-list* *E776* A location list is similar to a quickfix list and contains a list of positions in files. A location list is associated with a window and each window can have a separate location list. A location list can be associated with only one window. The location list is independent of the quickfix list. When a window with a location list is split, the new window gets a copy of the location list. When there are no references to a location list, the location list is destroyed. The following quickfix commands can be used. The location list commands are similar to the quickfix commands, replacing the 'c' prefix in the quickfix command with 'l'. *:cc* :cc[!] [nr] Display error [nr]. If [nr] is omitted, the same error is displayed again. Without [!] this doesn't work when jumping to another buffer, the current buffer has been changed, there is the only window for the buffer and both 'hidden' and 'autowrite' are off. When jumping to another buffer with [!] any changes to the current buffer are lost, unless 'hidden' is set or there is another window for this buffer. The 'switchbuf' settings are respected when jumping to a buffer. *:ll* :ll[!] [nr] Same as ":cc", except the location list for the current window is used instead of the quickfix list. *:cn* *:cnext* *E553* :[count]cn[ext][!] Display the [count] next error in the list that includes a file name. If there are no file names at all, go to the [count] next error. See |:cc| for [!] and 'switchbuf'. *:lne* *:lnext* :[count]lne[xt][!] Same as ":cnext", except the location list for the current window is used instead of the quickfix list. :[count]cN[ext][!] *:cp* *:cprevious* *:cN* *:cNext* :[count]cp[revious][!] Display the [count] previous error in the list that includes a file name. If there are no file names at all, go to the [count] previous error. See |:cc| for [!] and 'switchbuf'. :[count]lN[ext][!] *:lp* *:lprevious* *:lN* *:lNext* :[count]lp[revious][!] Same as ":cNext" and ":cprevious", except the location list for the current window is used instead of the quickfix list. *:cnf* *:cnfile* :[count]cnf[ile][!] Display the first error in the [count] next file in the list that includes a file name. If there are no file names at all or if there is no next file, go to the [count] next error. See |:cc| for [!] and 'switchbuf'. *:lnf* *:lnfile* :[count]lnf[ile][!] Same as ":cnfile", except the location list for the current window is used instead of the quickfix list. :[count]cNf[ile][!] *:cpf* *:cpfile* *:cNf* *:cNfile* :[count]cpf[ile][!] Display the last error in the [count] previous file in the list that includes a file name. If there are no file names at all or if there is no next file, go to the [count] previous error. See |:cc| for [!] and 'switchbuf'. :[count]lNf[ile][!] *:lpf* *:lpfile* *:lNf* *:lNfile* :[count]lpf[ile][!] Same as ":cNfile" and ":cpfile", except the location list for the current window is used instead of the quickfix list. *:crewind* *:cr* :cr[ewind][!] [nr] Display error [nr]. If [nr] is omitted, the FIRST error is displayed. See |:cc|. *:lrewind* *:lr* :lr[ewind][!] [nr] Same as ":crewind", except the location list for the current window is used instead of the quickfix list. *:cfirst* *:cfir* :cfir[st][!] [nr] Same as ":crewind". *:lfirst* *:lfir* :lfir[st][!] [nr] Same as ":lrewind". *:clast* *:cla* :cla[st][!] [nr] Display error [nr]. If [nr] is omitted, the LAST error is displayed. See |:cc|. *:llast* *:lla* :lla[st][!] [nr] Same as ":clast", except the location list for the current window is used instead of the quickfix list. *:cq* *:cquit* :cq[uit][!] Quit Vim with an error code, so that the compiler will not compile the same file again. WARNING: All changes in files are lost! Also when the [!] is not used. It works like ":qall!" |:qall|, except that Vim returns a non-zero exit code. *:cf* *:cfile* :cf[ile][!] [errorfile] Read the error file and jump to the first error. This is done automatically when Vim is started with the -q option. You can use this command when you keep Vim running while compiling. If you give the name of the errorfile, the 'errorfile' option will be set to [errorfile]. See |:cc| for [!]. *:lf* *:lfile* :lf[ile][!] [errorfile] Same as ":cfile", except the location list for the current window is used instead of the quickfix list. You can not use the -q command-line option to set the location list. :cg[etfile] [errorfile] *:cg* *:cgetfile* Read the error file. Just like ":cfile" but don't jump to the first error. :lg[etfile] [errorfile] *:lg* *:lgetfile* Same as ":cgetfile", except the location list for the current window is used instead of the quickfix list. *:caddf* *:caddfile* :caddf[ile] [errorfile] Read the error file and add the errors from the errorfile to the current quickfix list. If a quickfix list is not present, then a new list is created. *:laddf* *:laddfile* :laddf[ile] [errorfile] Same as ":caddfile", except the location list for the current window is used instead of the quickfix list. *:cb* *:cbuffer* *E681* :cb[uffer][!] [bufnr] Read the error list from the current buffer. When [bufnr] is given it must be the number of a loaded buffer. That buffer will then be used instead of the current buffer. A range can be specified for the lines to be used. Otherwise all lines in the buffer are used. See |:cc| for [!]. *:lb* *:lbuffer* :lb[uffer][!] [bufnr] Same as ":cbuffer", except the location list for the current window is used instead of the quickfix list. *:cgetb* *:cgetbuffer* :cgetb[uffer] [bufnr] Read the error list from the current buffer. Just like ":cbuffer" but don't jump to the first error. *:lgetb* *:lgetbuffer* :lgetb[uffer] [bufnr] Same as ":cgetbuffer", except the location list for the current window is used instead of the quickfix list. *:caddb* *:caddbuffer* :caddb[uffer] [bufnr] Read the error list from the current buffer and add the errors to the current quickfix list. If a quickfix list is not present, then a new list is created. Otherwise, same as ":cbuffer". *:laddb* *:laddbuffer* :laddb[uffer] [bufnr] Same as ":caddbuffer", except the location list for the current window is used instead of the quickfix list. *:cex* *:cexpr* *E777* :cex[pr][!] {expr} Create a quickfix list using the result of {expr} and jump to the first error. If {expr} is a String, then each new-line terminated line in the String is processed using 'errorformat' and the result is added to the quickfix list. If {expr} is a List, then each String item in the list is processed and added to the quickfix list. Non String items in the List are ignored. See |:cc| for [!]. Examples: > :cexpr system('grep -n xyz *') :cexpr getline(1, '$') < *:lex* *:lexpr* :lex[pr][!] {expr} Same as ":cexpr", except the location list for the current window is used instead of the quickfix list. *:cgete* *:cgetexpr* :cgete[xpr] {expr} Create a quickfix list using the result of {expr}. Just like ":cexpr", but don't jump to the first error. *:lgete* *:lgetexpr* :lgete[xpr] {expr} Same as ":cgetexpr", except the location list for the current window is used instead of the quickfix list. *:cad* *:caddexpr* :cad[dexpr] {expr} Evaluate {expr} and add the resulting lines to the current quickfix list. If a quickfix list is not present, then a new list is created. The current cursor position will not be changed. See |:cexpr| for more information. Example: > :g/mypattern/caddexpr expand("%") . ":" . line(".") . ":" . getline(".") < *:lad* *:laddexpr* :lad[dexpr] {expr} Same as ":caddexpr", except the location list for the current window is used instead of the quickfix list. *:cl* *:clist* :cl[ist] [from] [, [to]] List all errors that are valid |quickfix-valid|. If numbers [from] and/or [to] are given, the respective range of errors is listed. A negative number counts from the last error backwards, -1 being the last error. The 'switchbuf' settings are respected when jumping to a buffer. :cl[ist]! [from] [, [to]] List all errors. *:lli* *:llist* :lli[st] [from] [, [to]] Same as ":clist", except the location list for the current window is used instead of the quickfix list. :lli[st]! [from] [, [to]] List all the entries in the location list for the current window. If you insert or delete lines, mostly the correct error location is still found because hidden marks are used. Sometimes, when the mark has been deleted for some reason, the message "line changed" is shown to warn you that the error location may not be correct. If you quit Vim and start again the marks are lost and the error locations may not be correct anymore. If vim is built with |+autocmd| support, two autocommands are available for running commands before and after a quickfix command (':make', ':grep' and so on) is executed. See |QuickFixCmdPre| and |QuickFixCmdPost| for details. *QuickFixCmdPost-example* When 'encoding' differs from the locale, the error messages may have a different encoding from what Vim is using. To convert the messages you can use this code: > function QfMakeConv() let qflist = getqflist() for i in qflist let i.text = iconv(i.text, "cp936", "utf-8") endfor call setqflist(qflist) endfunction au QuickfixCmdPost make call QfMakeConv() ============================================================================= 2. The error window *quickfix-window* *:cope* *:copen* :cope[n] [height] Open a window to show the current list of errors. When [height] is given, the window becomes that high (if there is room). Otherwise the window is made ten lines high. The window will contain a special buffer, with 'buftype' equal to "quickfix". Don't change this! If there already is a quickfix window, it will be made the current window. It is not possible to open a second quickfix window. *:lop* *:lopen* :lop[en] [height] Open a window to show the location list for the current window. Works only when the location list for the current window is present. You can have more than one location window opened at a time. Otherwise, it acts the same as ":copen". *:ccl* *:cclose* :ccl[ose] Close the quickfix window. *:lcl* *:lclose* :lcl[ose] Close the window showing the location list for the current window. *:cw* *:cwindow* :cw[indow] [height] Open the quickfix window when there are recognized errors. If the window is already open and there are no recognized errors, close the window. *:lw* *:lwindow* :lw[indow] [height] Same as ":cwindow", except use the window showing the location list for the current window. Normally the quickfix window is at the bottom of the screen. If there are vertical splits, it's at the bottom of the rightmost column of windows. To make it always occupy the full width: > :botright cwindow You can move the window around with |window-moving| commands. For example, to move it to the top: CTRL-W K The 'winfixheight' option will be set, which means that the window will mostly keep its height, ignoring 'winheight' and 'equalalways'. You can change the height manually (e.g., by dragging the status line above it with the mouse). In the quickfix window, each line is one error. The line number is equal to the error number. You can use ":.cc" to jump to the error under the cursor. Hitting the <Enter> key or double-clicking the mouse on a line has the same effect. The file containing the error is opened in the window above the quickfix window. If there already is a window for that file, it is used instead. If the buffer in the used window has changed, and the error is in another file, jumping to the error will fail. You will first have to make sure the window contains a buffer which can be abandoned. *CTRL-W_<Enter>* *CTRL-W_<CR>* You can use CTRL-W <Enter> to open a new window and jump to the error there. When the quickfix window has been filled, two autocommand events are triggered. First the 'filetype' option is set to "qf", which triggers the FileType event. Then the BufReadPost event is triggered, using "quickfix" for the buffer name. This can be used to perform some action on the listed errors. Example: > au BufReadPost quickfix setlocal modifiable \ | silent exe 'g/^/s//\=line(".")." "/' \ | setlocal nomodifiable This prepends the line number to each line. Note the use of "\=" in the substitute string of the ":s" command, which is used to evaluate an expression. The BufWinEnter event is also triggered, again using "quickfix" for the buffer name. Note: Making changes in the quickfix window has no effect on the list of errors. 'modifiable' is off to avoid making changes. If you delete or insert lines anyway, the relation between the text and the error number is messed up. If you really want to do this, you could write the contents of the quickfix window to a file and use ":cfile" to have it parsed and used as the new error list. *location-list-window* The location list window displays the entries in a location list. When you open a location list window, it is created below the current window and displays the location list for the current window. The location list window is similar to the quickfix window, except that you can have more than one location list window open at a time. When you use a location list command in this window, the displayed location list is used. When you select a file from the location list window, the following steps are used to find a window to edit the file: 1. If a window with the location list displayed in the location list window is present, then the file is opened in that window. 2. If the above step fails and if the file is already opened in another window, then that window is used. 3. If the above step fails then an existing window showing a buffer with 'buftype' not set is used. 4. If the above step fails, then the file is edited in a new window. In all of the above cases, if the location list for the selected window is not yet set, then it is set to the location list displayed in the location list window. ============================================================================= 3. Using more than one list of errors *quickfix-error-lists* So far has been assumed that there is only one list of errors. Actually the ten last used lists are remembered. When starting a new list, the previous ones are automatically kept. Two commands can be used to access older error lists. They set one of the existing error lists as the current one. *:colder* *:col* *E380* :col[der] [count] Go to older error list. When [count] is given, do this [count] times. When already at the oldest error list, an error message is given. *:lolder* *:lol* :lol[der] [count] Same as ":colder", except use the location list for the current window instead of the quickfix list. *:cnewer* *:cnew* *E381* :cnew[er] [count] Go to newer error list. When [count] is given, do this [count] times. When already at the newest error list, an error message is given. *:lnewer* *:lnew* :lnew[er] [count] Same as ":cnewer", except use the location list for the current window instead of the quickfix list. When adding a new error list, it becomes the current list. When ":colder" has been used and ":make" or ":grep" is used to add a new error list, one newer list is overwritten. This is especially useful if you are browsing with ":grep" |grep|. If you want to keep the more recent error lists, use ":cnewer 99" first. ============================================================================= 4. Using :make *:make_makeprg* *:mak* *:make* :mak[e][!] [arguments] 1. If vim was built with |+autocmd|, all relevant |QuickFixCmdPre| autocommands are executed. 2. If the 'autowrite' option is on, write any changed buffers 3. An errorfile name is made from 'makeef'. If 'makeef' doesn't contain "##", and a file with this name already exists, it is deleted. 4. The program given with the 'makeprg' option is started (default "make") with the optional [arguments] and the output is saved in the errorfile (for Unix it is also echoed on the screen). 5. The errorfile is read using 'errorformat'. 6. If vim was built with |+autocmd|, all relevant |QuickFixCmdPost| autocommands are executed. See example below. 7. If [!] is not given the first error is jumped to. 8. The errorfile is deleted. 9. You can now move through the errors with commands like |:cnext| and |:cprevious|, see above. This command does not accept a comment, any " characters are considered part of the arguments. *:lmak* *:lmake* :lmak[e][!] [arguments] Same as ":make", except the location list for the current window is used instead of the quickfix list. The ":make" command executes the command given with the 'makeprg' option. This is done by passing the command to the shell given with the 'shell' option. This works almost like typing ":!{makeprg} [arguments] {shellpipe} {errorfile}". {makeprg} is the string given with the 'makeprg' option. Any command can be used, not just "make". Characters '%' and '#' are expanded as usual on a command-line. You can use "%<" to insert the current file name without extension, or "#<" to insert the alternate file name without extension, for example: > :set makeprg=make\ #<.o [arguments] is anything that is typed after ":make". {shellpipe} is the 'shellpipe' option. {errorfile} is the 'makeef' option, with ## replaced to make it unique. The placeholder "$*" can be used for the argument list in {makeprog} if the command needs some additional characters after its arguments. The $* is replaced then by all arguments. Example: > :set makeprg=latex\ \\\\nonstopmode\ \\\\input\\{$*} or simpler > :let &mp = 'latex \\nonstopmode \\input\{$*}' "$*" can be given multiple times, for example: > :set makeprg=gcc\ -o\ $*\ $* The 'shellpipe' option defaults to ">" for the Amiga, MS-DOS and Win32. This means that the output of the compiler is saved in a file and not shown on the screen directly. For Unix "| tee" is used. The compiler output is shown on the screen and saved in a file the same time. Depending on the shell used "|& tee" or "2>&1| tee" is the default, so stderr output will be included. If 'shellpipe' is empty, the {errorfile} part will be omitted. This is useful for compilers that write to an errorfile themselves (e.g., Manx's Amiga C). Using QuickFixCmdPost to fix the encoding ~ It may be that 'encoding' is set to an encoding that differs from the messages your build program produces. This example shows how to fix this after Vim has read the error messages: > function QfMakeConv() let qflist = getqflist() for i in qflist let i.text = iconv(i.text, "cp936", "utf-8") endfor call setqflist(qflist) endfunction au QuickfixCmdPost make call QfMakeConv() (Example by Faque Cheng) ============================================================================== 5. Using :vimgrep and :grep *grep* *lid* Vim has two ways to find matches for a pattern: Internal and external. The advantage of the internal grep is that it works on all systems and uses the powerful Vim search patterns. An external grep program can be used when the Vim grep does not do what you want. The internal method will be slower, because files are read into memory. The advantages are: - Line separators and encoding are automatically recognized, as if a file is being edited. - Uses Vim search patterns. Multi-line patterns can be used. - When plugins are enabled: compressed and remote files can be searched. |gzip| |netrw| To be able to do this Vim loads each file as if it is being edited. When there is no match in the file the associated buffer is wiped out again. The 'hidden' option is ignored here to avoid running out of memory or file descriptors when searching many files. However, when the |:hide| command modifier is used the buffers are kept loaded. This makes following searches in the same files a lot faster. 5.1 using Vim's internal grep *:vim* *:vimgrep* *E682* *E683* :vim[grep][!] /{pattern}/[g][j] {file} ... Search for {pattern} in the files {file} ... and set the error list to the matches. Without the 'g' flag each line is added only once. With 'g' every match is added. {pattern} is a Vim search pattern. Instead of enclosing it in / any non-ID character (see |'isident'|) can be used, so long as it does not appear in {pattern}. 'ignorecase' applies. To overrule it put |/\c| in the pattern to ignore case or |/\C| to match case. 'smartcase' is not used. When a number is put before the command this is used as the maximum number of matches to find. Use ":1vimgrep pattern file" to find only the first. Useful if you only want to check if there is a match and quit quickly when it's found. Without the 'j' flag Vim jumps to the first match. With 'j' only the quickfix list is updated. With the [!] any changes in the current buffer are abandoned. Every second or so the searched file name is displayed to give you an idea of the progress made. Examples: > :vimgrep /an error/ *.c :vimgrep /\<FileName\>/ *.h include/* :vimgrep /myfunc/ **/*.c < For the use of "**" see |starstar-wildcard|. :vim[grep][!] {pattern} {file} ... Like above, but instead of enclosing the pattern in a non-ID character use a white-separated pattern. The pattern must start with an ID character. Example: > :vimgrep Error *.c < *:lv* *:lvimgrep* :lv[imgrep][!] /{pattern}/[g][j] {file} ... :lv[imgrep][!] {pattern} {file} ... Same as ":vimgrep", except the location list for the current window is used instead of the quickfix list. *:vimgrepa* *:vimgrepadd* :vimgrepa[dd][!] /{pattern}/[g][j] {file} ... :vimgrepa[dd][!] {pattern} {file} ... Just like ":vimgrep", but instead of making a new list of errors the matches are appended to the current list. *:lvimgrepa* *:lvimgrepadd* :lvimgrepa[dd][!] /{pattern}/[g][j] {file} ... :lvimgrepa[dd][!] {pattern} {file} ... Same as ":vimgrepadd", except the location list for the current window is used instead of the quickfix list. 5.2 External grep Vim can interface with "grep" and grep-like programs (such as the GNU id-utils) in a similar way to its compiler integration (see |:make| above). [Unix trivia: The name for the Unix "grep" command comes from ":g/re/p", where "re" stands for Regular Expression.] *:gr* *:grep* :gr[ep][!] [arguments] Just like ":make", but use 'grepprg' instead of 'makeprg' and 'grepformat' instead of 'errorformat'. When 'grepprg' is "internal" this works like |:vimgrep|. Note that the pattern needs to be enclosed in separator characters then. *:lgr* *:lgrep* :lgr[ep][!] [arguments] Same as ":grep", except the location list for the current window is used instead of the quickfix list. *:grepa* *:grepadd* :grepa[dd][!] [arguments] Just like ":grep", but instead of making a new list of errors the matches are appended to the current list. Example: > :grep nothing % :bufdo grepadd! something % < The first command makes a new error list which is empty. The second command executes "grepadd" for each listed buffer. Note the use of ! to avoid that ":grepadd" jumps to the first error, which is not allowed with |:bufdo|. *:lgrepa* *:lgrepadd* :lgrepa[dd][!] [arguments] Same as ":grepadd", except the location list for the current window is used instead of the quickfix list. 5.3 Setting up external grep If you have a standard "grep" program installed, the :grep command may work well with the defaults. The syntax is very similar to the standard command: > :grep foo *.c Will search all files with the .c extension for the substring "foo". The arguments to :grep are passed straight to the "grep" program, so you can use whatever options your "grep" supports. By default, :grep invokes grep with the -n option (show file and line numbers). You can change this with the 'grepprg' option. You will need to set 'grepprg' if: a) You are using a program that isn't called "grep" b) You have to call grep with a full path c) You want to pass other options automatically (e.g. case insensitive search.) Once "grep" has executed, Vim parses the results using the 'grepformat' option. This option works in the same way as the 'errorformat' option - see that for details. You may need to change 'grepformat' from the default if your grep outputs in a non-standard format, or you are using some other program with a special format. Once the results are parsed, Vim loads the first file containing a match and jumps to the appropriate line, in the same way that it jumps to a compiler error in |quickfix| mode. You can then use the |:cnext|, |:clist|, etc. commands to see the other matches. 5.4 Using :grep with id-utils You can set up :grep to work with the GNU id-utils like this: > :set grepprg=lid\ -Rgrep\ -s :set grepformat=%f:%l:%m then > :grep (regexp) works just as you'd expect. (provided you remembered to mkid first :) 5.5 Browsing source code with :vimgrep or :grep Using the stack of error lists that Vim keeps, you can browse your files to look for functions and the functions they call. For example, suppose that you have to add an argument to the read_file() function. You enter this command: > :vimgrep /\<read_file\>/ *.c You use ":cn" to go along the list of matches and add the argument. At one place you have to get the new argument from a higher level function msg(), and need to change that one too. Thus you use: > :vimgrep /\<msg\>/ *.c While changing the msg() functions, you find another function that needs to get the argument from a higher level. You can again use ":vimgrep" to find these functions. Once you are finished with one function, you can use > :colder to go back to the previous one. This works like browsing a tree: ":vimgrep" goes one level deeper, creating a list of branches. ":colder" goes back to the previous level. You can mix this use of ":vimgrep" and "colder" to browse all the locations in a tree-like way. If you do this consistently, you will find all locations without the need to write down a "todo" list. ============================================================================= 6. Selecting a compiler *compiler-select* *:comp* *:compiler* *E666* :comp[iler][!] {name} Set options to work with compiler {name}. Without the "!" options are set for the current buffer. With "!" global options are set. If you use ":compiler foo" in "file.foo" and then ":compiler! bar" in another buffer, Vim will keep on using "foo" in "file.foo". {not available when compiled without the |+eval| feature} The Vim plugins in the "compiler" directory will set options to use the selected compiler. For ":compiler" local options are set, for ":compiler!" global options. *current_compiler* To support older Vim versions, the plugins always use "current_compiler" and not "b:current_compiler". What the command actually does is the following: - Delete the "current_compiler" and "b:current_compiler" variables. - Define the "CompilerSet" user command. With "!" it does ":set", without "!" it does ":setlocal". - Execute ":runtime! compiler/{name}.vim". The plugins are expected to set options with "CompilerSet" and set the "current_compiler" variable to the name of the compiler. - Delete the "CompilerSet" user command. - Set "b:current_compiler" to the value of "current_compiler". - Without "!" the old value of "current_compiler" is restored. For writing a compiler plugin, see |write-compiler-plugin|. GCC *quickfix-gcc* *compiler-gcc* There's one variable you can set for the GCC compiler: g:compiler_gcc_ignore_unmatched_lines Ignore lines that don't match any patterns defined for GCC. Useful if output from commands run from make are generating false positives. MANX AZTEC C *quickfix-manx* *compiler-manx* To use Vim with Manx's Aztec C compiler on the Amiga you should do the following: - Set the CCEDIT environment variable with the command: > mset "CCEDIT=vim -q" - Compile with the -qf option. If the compiler finds any errors, Vim is started and the cursor is positioned on the first error. The error message will be displayed on the last line. You can go to other errors with the commands mentioned above. You can fix the errors and write the file(s). - If you exit Vim normally the compiler will re-compile the same file. If you exit with the :cq command, the compiler will terminate. Do this if you cannot fix the error, or if another file needs to be compiled first. There are some restrictions to the Quickfix mode on the Amiga. The compiler only writes the first 25 errors to the errorfile (Manx's documentation does not say how to get more). If you want to find the others, you will have to fix a few errors and exit the editor. After recompiling, up to 25 remaining errors will be found. If Vim was started from the compiler, the :sh and some :! commands will not work, because Vim is then running in the same process as the compiler and stdin (standard input) will not be interactive. PERL *quickfix-perl* *compiler-perl* The Perl compiler plugin doesn't actually compile, but invokes Perl's internal syntax checking feature and parses the output for possible errors so you can correct them in quick-fix mode. Warnings are forced regardless of "no warnings" or "$^W = 0" within the file being checked. To disable this set g:perl_compiler_force_warnings to a zero value. For example: > let g:perl_compiler_force_warnings = 0 PYUNIT COMPILER *compiler-pyunit* This is not actually a compiler, but a unit testing framework for the Python language. It is included into standard Python distribution starting from version 2.0. For older versions, you can get it from http://pyunit.sourceforge.net. When you run your tests with the help of the framework, possible errors are parsed by Vim and presented for you in quick-fix mode. Unfortunately, there is no standard way to run the tests. The alltests.py script seems to be used quite often, that's all. Useful values for the 'makeprg' options therefore are: setlocal makeprg=./alltests.py " Run a testsuite setlocal makeprg=python % " Run a single testcase Also see http://vim.sourceforge.net/tip_view.php?tip_id=280. TEX COMPILER *compiler-tex* Included in the distribution compiler for TeX ($VIMRUNTIME/compiler/tex.vim) uses make command if possible. If the compiler finds a file named "Makefile" or "makefile" in the current directory, it supposes that you want to process your *TeX files with make, and the makefile does the right work. In this case compiler sets 'errorformat' for *TeX output and leaves 'makeprg' untouched. If neither "Makefile" nor "makefile" is found, the compiler will not use make. You can force the compiler to ignore makefiles by defining b:tex_ignore_makefile or g:tex_ignore_makefile variable (they are checked for existence only). If the compiler chose not to use make, it need to choose a right program for processing your input. If b:tex_flavor or g:tex_flavor (in this precedence) variable exists, it defines TeX flavor for :make (actually, this is the name of executed command), and if both variables do not exist, it defaults to "latex". For example, while editing chapter2.tex \input-ed from mypaper.tex written in AMS-TeX: > :let b:tex_flavor = 'amstex' :compiler tex < [editing...] > :make mypaper Note that you must specify a name of the file to process as an argument (to process the right file when editing \input-ed or \include-ed file; portable solution for substituting % for no arguments is welcome). This is not in the semantics of make, where you specify a target, not source, but you may specify filename without extension ".tex" and mean this as "make filename.dvi or filename.pdf or filename.some_result_extension according to compiler". Note: tex command line syntax is set to usable both for MikTeX (suggestion by Srinath Avadhanula) and teTeX (checked by Artem Chuprina). Suggestion from |errorformat-LaTeX| is too complex to keep it working for different shells and OSes and also does not allow to use other available TeX options, if any. If your TeX doesn't support "-interaction=nonstopmode", please report it with different means to express \nonstopmode from the command line. ============================================================================= 7. The error format *error-file-format* *errorformat* *E372* *E373* *E374* *E375* *E376* *E377* *E378* The 'errorformat' option specifies a list of formats that are recognized. The first format that matches with an error message is used. You can add several formats for different messages your compiler produces, or even entries for multiple compilers. See |efm-entries|. Each entry in 'errorformat' is a scanf-like string that describes the format. First, you need to know how scanf works. Look in the documentation of your C compiler. Below you find the % items that Vim understands. Others are invalid. Special characters in 'errorformat' are comma and backslash. See |efm-entries| for how to deal with them. Note that a literal "%" is matched by "%%", thus it is not escaped with a backslash. Note: By default the difference between upper and lowercase is ignored. If you want to match case, add "\C" to the pattern |/\C|. Basic items %f file name (finds a string) %l line number (finds a number) %c column number (finds a number representing character column of the error, (1 <tab> == 1 character column)) %v virtual column number (finds a number representing screen column of the error (1 <tab> == 8 screen columns)) %t error type (finds a single character) %n error number (finds a number) %m error message (finds a string) %r matches the "rest" of a single-line file message %O/P/Q %p pointer line (finds a sequence of '-', '.' or ' ' and uses the length for the column number) %*{conv} any scanf non-assignable conversion %% the single '%' character %s search text (finds a string) The "%f" conversion may depend on the current 'isfname' setting. "~/" is expanded to the home directory and environment variables are expanded. The "%f" and "%m" conversions have to detect the end of the string. This normally happens by matching following characters and items. When nothing is following the rest of the line is matched. If "%f" is followed by a '%' or a backslash, it will look for a sequence of 'isfname' characters. On MS-DOS, MS-Windows and OS/2 a leading "C:" will be included in "%f", even when using "%f:". This means that a file name which is a single alphabetical letter will not be detected. The "%p" conversion is normally followed by a "^". It's used for compilers that output a line like: > ^ or > ---------^ to indicate the column of the error. This is to be used in a multi-line error message. See |errorformat-javac| for a useful example. The "%s" conversion specifies the text to search for to locate the error line. The text is used as a literal string. The anchors "^" and "$" are added to the text to locate the error line exactly matching the search text and the text is prefixed with the "\V" atom to make it "very nomagic". The "%s" conversion can be used to locate lines without a line number in the error output. Like the output of the "grep" shell command. When the pattern is present the line number will not be used. Changing directory The following uppercase conversion characters specify the type of special format strings. At most one of them may be given as a prefix at the begin of a single comma-separated format pattern. Some compilers produce messages that consist of directory names that have to be prepended to each file name read by %f (example: GNU make). The following codes can be used to scan these directory names; they will be stored in an internal directory stack. *E379* %D "enter directory" format string; expects a following %f that finds the directory name %X "leave directory" format string; expects following %f When defining an "enter directory" or "leave directory" format, the "%D" or "%X" has to be given at the start of that substring. Vim tracks the directory changes and prepends the current directory to each erroneous file found with a relative path. See |quickfix-directory-stack| for details, tips and limitations. Multi-line messages *errorformat-multi-line* It is possible to read the output of programs that produce multi-line messages, i.e. error strings that consume more than one line. Possible prefixes are: %E start of a multi-line error message %W start of a multi-line warning message %I start of a multi-line informational message %A start of a multi-line message (unspecified type) %> for next line start with current pattern again |efm-%>| %C continuation of a multi-line message %Z end of a multi-line message These can be used with '+' and '-', see |efm-ignore| below. Using "\n" in the pattern won't work to match multi-line messages. Example: Your compiler happens to write out errors in the following format (leading line numbers not being part of the actual output): 1 Error 275 ~ 2 line 42 ~ 3 column 3 ~ 4 ' ' expected after '--' ~ The appropriate error format string has to look like this: > :set efm=%EError\ %n,%Cline\ %l,%Ccolumn\ %c,%Z%m And the |:clist| error message generated for this error is: 1:42 col 3 error 275: ' ' expected after '--' Another example: Think of a Python interpreter that produces the following error message (line numbers are not part of the actual output): 1 ============================================================== 2 FAIL: testGetTypeIdCachesResult (dbfacadeTest.DjsDBFacadeTest) 3 -------------------------------------------------------------- 4 Traceback (most recent call last): 5 File "unittests/dbfacadeTest.py", line 89, in testFoo 6 self.assertEquals(34, dtid) 7 File "/usr/lib/python2.2/unittest.py", line 286, in 8 failUnlessEqual 9 raise self.failureException, \ 10 AssertionError: 34 != 33 11 12 -------------------------------------------------------------- 13 Ran 27 tests in 0.063s Say you want |:clist| write the relevant information of this message only, namely: 5 unittests/dbfacadeTest.py:89: AssertionError: 34 != 33 Then the error format string could be defined as follows: > :set efm=%C\ %.%#,%A\ \ File\ \"%f\"\\,\ line\ %l%.%#,%Z%[%^\ ]%\\@=%m Note that the %C string is given before the %A here: since the expression ' %.%#' (which stands for the regular expression ' .*') matches every line starting with a space, followed by any characters to the end of the line, it also hides line 7 which would trigger a separate error message otherwise. Error format strings are always parsed pattern by pattern until the first match occurs. *efm-%>* The %> item can be used to avoid trying patterns that appear earlier in 'errorformat'. This is useful for patterns that match just about anything. For example, if the error looks like this: Error in line 123 of foo.c: ~ unknown variable "i" ~ This can be found with: > :set efm=xxx,%E%>Error in line %l of %f:,%Z%m Where "xxx" has a pattern that would also match the second line. Important: There is no memory of what part of the errorformat matched before; every line in the error file gets a complete new run through the error format lines. For example, if one has: > setlocal efm=aa,bb,cc,dd,ee Where aa, bb, etc. are error format strings. Each line of the error file will be matched to the pattern aa, then bb, then cc, etc. Just because cc matched the previous error line does _not_ mean that dd will be tried first on the current line, even if cc and dd are multi-line errorformat strings. Separate file name *errorformat-separate-filename* These prefixes are useful if the file name is given once and multiple messages follow that refer to this file name. %O single-line file message: overread the matched part %P single-line file message: push file %f onto the stack %Q single-line file message: pop the last file from stack Example: Given a compiler that produces the following error logfile (without leading line numbers): 1 [a1.tt] 2 (1,17) error: ';' missing 3 (21,2) warning: variable 'z' not defined 4 (67,3) error: end of file found before string ended 5 6 [a2.tt] 7 8 [a3.tt] 9 NEW compiler v1.1 10 (2,2) warning: variable 'x' not defined 11 (67,3) warning: 's' already defined This logfile lists several messages for each file enclosed in [...] which are properly parsed by an error format like this: > :set efm=%+P[%f],(%l\\,%c)%*[\ ]%t%*[^:]:\ %m,%-Q A call of |:clist| writes them accordingly with their correct filenames: 2 a1.tt:1 col 17 error: ';' missing 3 a1.tt:21 col 2 warning: variable 'z' not defined 4 a1.tt:67 col 3 error: end of file found before string ended 8 a3.tt:2 col 2 warning: variable 'x' not defined 9 a3.tt:67 col 3 warning: 's' already defined Unlike the other prefixes that all match against whole lines, %P, %Q and %O can be used to match several patterns in the same line. Thus it is possible to parse even nested files like in the following line: {"file1" {"file2" error1} error2 {"file3" error3 {"file4" error4 error5}}} The %O then parses over strings that do not contain any push/pop file name information. See |errorformat-LaTeX| for an extended example. Ignoring and using whole messages *efm-ignore* The codes '+' or '-' can be combined with the uppercase codes above; in that case they have to precede the letter, e.g. '%+A' or '%-G': %- do not include the matching multi-line in any output %+ include the whole matching line in the %m error string One prefix is only useful in combination with '+' or '-', namely %G. It parses over lines containing general information like compiler version strings or other headers that can be skipped. %-G ignore this message %+G general message Pattern matching The scanf()-like "%*[]" notation is supported for backward-compatibility with previous versions of Vim. However, it is also possible to specify (nearly) any Vim supported regular expression in format strings. Since meta characters of the regular expression language can be part of ordinary matching strings or file names (and therefore internally have to be escaped), meta symbols have to be written with leading '%': %\ The single '\' character. Note that this has to be escaped ("%\\") in ":set errorformat=" definitions. %. The single '.' character. %# The single '*'(!) character. %^ The single '^' character. Note that this is not useful, the pattern already matches start of line. %$ The single '$' character. Note that this is not useful, the pattern already matches end of line. %[ The single '[' character for a [] character range. %~ The single '~' character. When using character classes in expressions (see |/\i| for an overview), terms containing the "\+" quantifier can be written in the scanf() "%*" notation. Example: "%\\d%\\+" ("\d\+", "any number") is equivalent to "%*\\d". Important note: The \(...\) grouping of sub-matches can not be used in format specifications because it is reserved for internal conversions. Multiple entries in 'errorformat' *efm-entries* To be able to detect output from several compilers, several format patterns may be put in 'errorformat', separated by commas (note: blanks after the comma are ignored). The first pattern that has a complete match is used. If no match is found, matching parts from the last one will be used, although the file name is removed and the error message is set to the whole message. If there is a pattern that may match output from several compilers (but not in a right way), put it after one that is more restrictive. To include a comma in a pattern precede it with a backslash (you have to type two in a ":set" command). To include a backslash itself give two backslashes (you have to type four in a ":set" command). You also need to put a backslash before a space for ":set". Valid matches *quickfix-valid* If a line does not completely match one of the entries in 'errorformat', the whole line is put in the error message and the entry is marked "not valid" These lines are skipped with the ":cn" and ":cp" commands (unless there is no valid line at all). You can use ":cl!" to display all the error messages. If the error format does not contain a file name Vim cannot switch to the correct file. You will have to do this by hand. Examples The format of the file from the Amiga Aztec compiler is: filename>linenumber:columnnumber:errortype:errornumber:errormessage filename name of the file in which the error was detected linenumber line number where the error was detected columnnumber column number where the error was detected errortype type of the error, normally a single 'E' or 'W' errornumber number of the error (for lookup in the manual) errormessage description of the error This can be matched with this 'errorformat' entry: %f>%l:%c:%t:%n:%m Some examples for C compilers that produce single-line error outputs: %f:%l:\ %t%*[^0123456789]%n:\ %m for Manx/Aztec C error messages (scanf() doesn't understand [0-9]) %f\ %l\ %t%*[^0-9]%n:\ %m for SAS C \"%f\"\\,%*[^0-9]%l:\ %m for generic C compilers %f:%l:\ %m for GCC %f:%l:\ %m,%Dgmake[%*\\d]:\ Entering\ directory\ `%f', %Dgmake[%*\\d]:\ Leaving\ directory\ `%f' for GCC with gmake (concat the lines!) %f(%l)\ :\ %*[^:]:\ %m old SCO C compiler (pre-OS5) %f(%l)\ :\ %t%*[^0-9]%n:\ %m idem, with error type and number %f:%l:\ %m,In\ file\ included\ from\ %f:%l:,\^I\^Ifrom\ %f:%l%m for GCC, with some extras Extended examples for the handling of multi-line messages are given below, see |errorformat-Jikes| and |errorformat-LaTeX|. Note the backslash in front of a space and double quote. It is required for the :set command. There are two backslashes in front of a comma, one for the :set command and one to avoid recognizing the comma as a separator of error formats. Filtering messages If you have a compiler that produces error messages that do not fit in the format string, you could write a program that translates the error messages into this format. You can use this program with the ":make" command by changing the 'makeprg' option. For example: > :set mp=make\ \\\|&\ error_filter The backslashes before the pipe character are required to avoid it to be recognized as a command separator. The backslash before each space is required for the set command. ============================================================================= 8. The directory stack *quickfix-directory-stack* Quickfix maintains a stack for saving all used directories parsed from the make output. For GNU-make this is rather simple, as it always prints the absolute path of all directories it enters and leaves. Regardless if this is done via a 'cd' command in the makefile or with the parameter "-C dir" (change to directory before reading the makefile). It may be useful to use the switch "-w" to force GNU-make to print out the working directory before and after processing. Maintaining the correct directory is more complicated if you don't use GNU-make. AIX-make for example doesn't print any information about its working directory. Then you need to enhance the makefile. In the makefile of LessTif there is a command which echoes "Making {target} in {dir}". The special problem here is that it doesn't print informations on leaving the directory and that it doesn't print the absolute path. To solve the problem with relative paths and missing "leave directory" messages Vim uses following algorithm: 1) Check if the given directory is a subdirectory of the current directory. If this is true, store it as the current directory. 2) If it is not a subdir of the current directory, try if this is a subdirectory of one of the upper directories. 3) If the directory still isn't found, it is assumed to be a subdirectory of Vim's current directory. Additionally it is checked for every file, if it really exists in the identified directory. If not, it is searched in all other directories of the directory stack (NOT the directory subtree!). If it is still not found, it is assumed that it is in Vim's current directory. There are limitation in this algorithm. This examples assume that make just prints information about entering a directory in the form "Making all in dir". 1) Assume you have following directories and files: ./dir1 ./dir1/file1.c ./file1.c If make processes the directory "./dir1" before the current directory and there is an error in the file "./file1.c", you will end up with the file "./dir1/file.c" loaded by Vim. This can only be solved with a "leave directory" message. 2) Assume you have following directories and files: ./dir1 ./dir1/dir2 ./dir2 You get the following: Make output Directory interpreted by Vim ------------------------ ---------------------------- Making all in dir1 ./dir1 Making all in dir2 ./dir1/dir2 Making all in dir2 ./dir1/dir2 This can be solved by printing absolute directories in the "enter directory" message or by printing "leave directory" messages.. To avoid this problems, ensure to print absolute directory names and "leave directory" messages. Examples for Makefiles: Unix: libs: for dn in $(LIBDIRS); do \ (cd $$dn; echo "Entering dir '$$(pwd)'"; make); \ echo "Leaving dir"; \ done Add %DEntering\ dir\ '%f',%XLeaving\ dir to your 'errorformat' to handle the above output. Note that Vim doesn't check if the directory name in a "leave directory" messages is the current directory. This is why you could just use the message "Leaving dir". ============================================================================= 9. Specific error file formats *errorformats* *errorformat-Jikes* Jikes(TM), a source-to-bytecode Java compiler published by IBM Research, produces simple multi-line error messages. An 'errorformat' string matching the produced messages is shown below. The following lines can be placed in the user's |vimrc| to overwrite Vim's recognized default formats, or see |:set+=| how to install this format additionally to the default. > :set efm=%A%f:%l:%c:%*\\d:%*\\d:, \%C%*\\s%trror:%m, \%+C%*[^:]%trror:%m, \%C%*\\s%tarning:%m, \%C%m < Jikes(TM) produces a single-line error message when invoked with the option "+E", and can be matched with the following: > :setl efm=%f:%l:%v:%*\\d:%*\\d:%*\\s%m < *errorformat-javac* This 'errorformat' has been reported to work well for javac, which outputs a line with "^" to indicate the column of the error: > :setl efm=%A%f:%l:\ %m,%-Z%p^,%-C%.%# or: > :setl efm=%A%f:%l:\ %m,%+Z%p^,%+C%.%#,%-G%.%# < Here is an alternative from Michael F. Lamb for Unix that filters the errors first: > :setl errorformat=%Z%f:%l:\ %m,%A%p^,%-G%*[^sl]%.%# :setl makeprg=javac\ %\ 2>&1\ \\\|\ vim-javac-filter You need to put the following in "vim-javac-filter" somewhere in your path (e.g., in ~/bin) and make it executable: > #!/bin/sed -f /\^$/s/\t/\ /g;/:[0-9]\+:/{h;d};/^[ \t]*\^/G; In English, that sed script: - Changes single tabs to single spaces and - Moves the line with the filename, line number, error message to just after the pointer line. That way, the unused error text between doesn't break vim's notion of a "multi-line message" and also doesn't force us to include it as a "continuation of a multi-line message." *errorformat-ant* For ant (http://jakarta.apache.org/) the above errorformat has to be modified to honour the leading [javac] in front of each javac output line: > :set efm=%A\ %#[javac]\ %f:%l:\ %m,%-Z\ %#[javac]\ %p^,%-C%.%# The 'errorformat' can also be configured to handle ant together with either javac or jikes. If you're using jikes, you should tell ant to use jikes' +E command line switch which forces jikes to generate one-line error messages. This is what the second line (of a build.xml file) below does: > <property name = "build.compiler" value = "jikes"/> <property name = "build.compiler.emacs" value = "true"/> The 'errorformat' which handles ant with both javac and jikes is: > :set efm=\ %#[javac]\ %#%f:%l:%c:%*\\d:%*\\d:\ %t%[%^:]%#:%m, \%A\ %#[javac]\ %f:%l:\ %m,%-Z\ %#[javac]\ %p^,%-C%.%# < *errorformat-jade* parsing jade (see http://www.jclark.com/) errors is simple: > :set efm=jade:%f:%l:%c:%t:%m < *errorformat-LaTeX* The following is an example how an 'errorformat' string can be specified for the (La)TeX typesetting system which displays error messages over multiple lines. The output of ":clist" and ":cc" etc. commands displays multi-lines in a single line, leading white space is removed. It should be easy to adopt the above LaTeX errorformat to any compiler output consisting of multi-line errors. The commands can be placed in a |vimrc| file or some other Vim script file, e.g. a script containing LaTeX related stuff which is loaded only when editing LaTeX sources. Make sure to copy all lines of the example (in the given order), afterwards remove the comment lines. For the '\' notation at the start of some lines see |line-continuation|. First prepare 'makeprg' such that LaTeX will report multiple errors; do not stop when the first error has occurred: > :set makeprg=latex\ \\\\nonstopmode\ \\\\input\\{$*} < Start of multi-line error messages: > :set efm=%E!\ LaTeX\ %trror:\ %m, \%E!\ %m, < Start of multi-line warning messages; the first two also include the line number. Meaning of some regular expressions: - "%.%#" (".*") matches a (possibly empty) string - "%*\\d" ("\d\+") matches a number > \%+WLaTeX\ %.%#Warning:\ %.%#line\ %l%.%#, \%+W%.%#\ at\ lines\ %l--%*\\d, \%WLaTeX\ %.%#Warning:\ %m, < Possible continuations of error/warning messages; the first one also includes the line number: > \%Cl.%l\ %m, \%+C\ \ %m., \%+C%.%#-%.%#, \%+C%.%#[]%.%#, \%+C[]%.%#, \%+C%.%#%[{}\\]%.%#, \%+C<%.%#>%.%#, \%C\ \ %m, < Lines that match the following patterns do not contain any important information; do not include them in messages: > \%-GSee\ the\ LaTeX%m, \%-GType\ \ H\ <return>%m, \%-G\ ...%.%#, \%-G%.%#\ (C)\ %.%#, \%-G(see\ the\ transcript%.%#), < Generally exclude any empty or whitespace-only line from being displayed: > \%-G\\s%#, < The LaTeX output log does not specify the names of erroneous source files per line; rather they are given globally, enclosed in parentheses. The following patterns try to match these names and store them in an internal stack. The patterns possibly scan over the same input line (one after another), the trailing "%r" conversion indicates the "rest" of the line that will be parsed in the next go until the end of line is reached. Overread a file name enclosed in '('...')'; do not push it on a stack since the file apparently does not contain any error: > \%+O(%f)%r, < Push a file name onto the stack. The name is given after '(': > \%+P(%f%r, \%+P\ %\\=(%f%r, \%+P%*[^()](%f%r, \%+P[%\\d%[^()]%#(%f%r, < Pop the last stored file name when a ')' is scanned: > \%+Q)%r, \%+Q%*[^()])%r, \%+Q[%\\d%*[^()])%r Note that in some cases file names in the LaTeX output log cannot be parsed properly. The parser might have been messed up by unbalanced parentheses then. The above example tries to catch the most relevant cases only. You can customize the given setting to suit your own purposes, for example, all the annoying "Overfull ..." warnings could be excluded from being recognized as an error. Alternatively to filtering the LaTeX compiler output, it is also possible to directly read the *.log file that is produced by the [La]TeX compiler. This contains even more useful information about possible error causes. However, to properly parse such a complex file, an external filter should be used. See the description further above how to make such a filter known by Vim. *errorformat-Perl* In $VIMRUNTIME/tools you can find the efm_perl.pl script, which filters Perl error messages into a format that quickfix mode will understand. See the start of the file about how to use it. (This script is deprecated, see |compiler-perl|.) vim:tw=78:ts=8:ft=help:norl: