a new feature in emacs 23

perm url: http://xahlee.org/emacs/modernization_find_replace_multi.html

Emacs Suggestion: dired-do-query-replace-regex Replace ALL

Xah Lee, 2008-05

[Note: The following is a suggestion sent to FSF official emacs bug channel, on 2008-05-17, as bug 267. This suggestion is implemented in emacs 23.1, released on 2009-07. See New Features in Emacs 23.]

Emacs suggestion:

when in dired mode in the processing of doing dired-do-query-replace- regexp (shortcut Q), it offers the ability to do replacement without asking on the current file, by pressing the ! key, but it would be nice to have:

  • (1) the ability to do ALL replacements on ALL files without further asking.
  • (2) the ability to not do any replacement on the current file.

I use dired-do-query-replace-regexp few times a week, typically on tens of files, but maybe once in a month i do it over hundreds or thousands of files on a website.

Often, after some replacement, it becomes obvious that it is safe to do ALL replacement for ALL files. When this is needed, typically i can just hold down the ! key. But when the number of files is hundreds or more, typically there will be several 5 second delays where i assume emacs is doing garbage collection. The bottom line is that, holding down ! is not a good solution because i end up having to wait a total of maybe 20 or 30 seconds during the whole operation.

For this reason, i often abort the operation, then switch to a perl or python script i've written that does blind find/replace on all files. This will finish the job on thousands of files within 5 seconds. But I couldn't reply on this since often i need interactively see and confirm the replacements, which emacs provides. (emacs also lets me view the change file afterward or decide to save them... etc, which i won't be able to do with perl...)

So, it would be nice, if emacs's dired-do-query-replace-regexp offer the option to do all replacement on all files.

Perhaps this can be introduced as pressing the @ key.

For suggestion item (2), the scenario is that sometimes i have a big file and it is apparent that i don't want any replacement to happen on that file. As things are, i have to repeatedly pressing n while paying attention not to go over to the next file. This is a tiring process.

dired-do-query-replace-regexp is very nice and i've been using it for 2 years. For just tens of files the above isn't a problem. But when doing few hundred files, and on tens of replacement pairs (i.e. using dired-do-query-replace-regexp tens of times, each time on hundreds of files), the above features seems a very good one to have. It seems also a logical feature complementing the various recursive edit and edit-replacement-text currently provided by emacs.

PS some background... I'm doing on static websites, which has few hundred or thousand files, such as systematically changing a style markup, etc. In the past 2 days, i spent entire 2 days using dired-do- query-replace-regexp (in combination with perl and python script).


Popular posts from this blog

11 Years of Writing About Emacs

does md5 creates more randomness?

Google Code shutting down, future of ErgoEmacs