Re: [O] Web site bug

[Achim Gratz]

Bastien writes:
> Can you point at an actual reproducible and simple bug with
> the current way Org defines autoloads?

After the introduction of org-loaddefs, the autoloads should be
extracted into two files and only the first is supposed to be loaded
before org is actually used and the second, org-loaddefs, is supposed to
be loaded by Org itself.  But when using a standalone Org, that first
file doesn't exist and there's no mechanism for the generated autoloads
(all in org-loaddefs) to be loaded before Org is used.

You suggest that in a standalone installation the user doesn't need to
do anything by way of customization: that's a bug, because the autoloads
in org-loaddefs for the standalone Org are not present _at all_ in that
situation.  That there may be some autoloads present from another Org
version (from Emacs) that by accident may point to a file that will then
cause Org to be loaded anyway doesn't make this bug go away.

Standalone Org needs a file separate from org-loaddefs to collect the
autoload definitions into that will be loaded on startup.  For Emacs,
this is loaddefs; for package manager, this is org-autoloads; for Org
standalone, I posit that we should keep org-install.  If we keep a
different system of how to handle autoloads depending on the type of
installation, then it will be impossible to test that everything works
correctly[1], not to mention the confusion that this discussion so amply
demonstrates.


[1] For instance, any first-level autoloaded function that somehow calls
into these files:

ob, ob-keys, ob-lob, ob-tangle, org-archive, org-ascii, org-attach,
org-bbdb, org-clock, org-datetree, org-docbook, org-element, org-exp,
org-feed, org-footnote, org-freemind, org-html, org-icalendar, org-id,
org-indent, org-irc, org-latex, org-lparse, org-mobile, org-odt,
org-plot, org-publish, org-remember, org-table, org-taskjuggler,
org-timer, org-xoxo

needs to make sure that org-loaddefs has been loaded, which currently
only happens in org, but I have not been able to ascertain that each
codepath to one of these functions will actually go through it.


Regards,
Achim.
-- 
+<[Q+ Matrix-12 WAVE#46+305 Neuron microQkb Andromeda XTk Blofeld]>+

Factory and User Sound Singles for Waldorf Blofeld:
http://Synth.Stromeko.net/Downloads.html#WaldorfSounds