dpm() and friends should work as expected during a Drush request

Looks pretty handy to me. I'll wait a bit for feedback before committing

Should we use drupal_is_cli() rather than a custom devel_is_drush()?

devel_set_message() will still handle message if drush isn't being used though right? And cron should be running as anonymous so messges are destroyed anyway.

Yeah, I'd rather not depend on drupal_is_cli() since this would be nice to have in d6. Although devel_set_message isn't in the 6 branch of devel anyhow, so that would need to be backported.

So the solution is we add a devel_is_cli() in d6 and use that instead just like backporting devel_set_message()? Not sure which approach would be better.

I see a couple problems with this approach. devel_set_message() is calling drush_log($msg, NULL), which doesn't print out without the debug flag. Without the second arg to drush_log, it still only prints if you pass verbose to drush.

I'd like to propose a considerably simpler approach:

function dpm($input, $name = NULL) {
  if (function_exists('drush_print_r')) {
    drush_print_r($input);
  }
  elseif (user_access('access devel information')) {
    $export = kprint_r($input, TRUE, $name);
    drupal_set_message($export);
  }
}

This avoids the issues discussed previously and it doesn't print html encoded characters on the cli.

Tagging

This patch needed a reroll, so I rerolled it.

If no-one else has any comments then in terms of functionality this works as expected.

Oops, this actually needs re-rolling against the latest release.

-  if (user_access('access devel information')) {
+  if (user_access('access devel information') || drupal_is_cli()) {

This looks wrong to me. Drush requests happen in the context of a user when -u is provided. Folks can pass in a user and then they will pass a user_access() test.

Not positive, but I think our move to Kint will fix this. dsm() and such now use Kint of that submodule is enabled so you can try for yourself and reopen if needed.