Re: [O] orgstuct++ does not lurk silently in the shadow
Hi Eric, Eric S Fraga e.fr...@ucl.ac.uk writes: If you're certain you're loading Org correctly, please try the attached patch and just tell me if the error disappears. The patch works! I'm still resisting applying this patch, because I don't understand why the infinite recursion occurs. Can you help me again by trying to give me a minimal setup where this infinite recursion occurs with latest Org? We're nearly there... thanks a lot! -- Bastien
Re: [O] orgstuct++ does not lurk silently in the shadow
Bastien b...@gnu.org writes: Hi Eric, Eric S Fraga e.fr...@ucl.ac.uk writes: If you're certain you're loading Org correctly, please try the attached patch and just tell me if the error disappears. The patch works! I'm still resisting applying this patch, because I don't understand why the infinite recursion occurs. Can you help me again by trying to give me a minimal setup where this infinite recursion occurs with latest Org? We're nearly there... thanks a lot! Strange. With orgstruct++-mode now enabled (manually, not through a hook), without your patch and without most of my org customisations, I cannot seem to get the infinite recursion happening. This is with org from git as of a few minutes ago. However, I cannot guarantee that some bits of org as shipped with emacs 24.1.50.1 haven't been loaded before I manually load the latest version. The problem is that I cannot get gnus to startup properly with emacs -Q. My gnus configuration is just too confused... :( Argg. In any case, given that it seems to work just fine without my customisations, there is something in my normal configuration (and Nicolas's, I guess) that would seem to be triggering the infinite recursion. I will start bisecting my org configuration to track this down but this could take some time! I would definitely suggest not applying that patch as it doesn't appear necessary in normal circumstances. Thanks, eric -- : Eric S Fraga (GnuPG: 0xC89193D8FFFCF67D) in Emacs 24.1.50.1 : using Org release_7.8.09-544-g505cc7
Re: [O] orgstuct++ does not lurk silently in the shadow
Hi Nicolas, Nicolas Goaziou n.goaz...@gmail.com writes: ((adaptive-fill-function (quote nil)) (adaptive-fill-regexp (quote \(\([ [...] This looks good. It's the same as the previous bug: when I try to debug `org-indent-line-function', the problem disappears. I pushed another fix: I now rely on the existing org-let function, which safely evaluate let constructs. -- Bastien
Re: [O] orgstuct++ does not lurk silently in the shadow
Bastien b...@gnu.org writes: Hi Nicolas, Nicolas Goaziou n.goaz...@gmail.com writes: ((adaptive-fill-function (quote nil)) (adaptive-fill-regexp (quote \(\([ [...] This looks good. It's the same as the previous bug: when I try to debug `org-indent-line-function', the problem disappears. I pushed another fix: I now rely on the existing org-let function, which safely evaluate let constructs. Bastien, unfortunately, I still have problems. I am using org completely up to date (a few minutes ago). I do not turn orgstruct++-mode on automatically at all any more but instead turn it on manually by M-x orgstruct++-mode RET once I'm in a Message buffer. I tried it just now, replying to your email as my test. Turned on the mode, tried to fill the quoted text and it all got mixed up. Undid that. Then went to the bottom and tried to insert a new line and got: org-indent-line-function: Lisp nesting exceeds `max-lisp-eval-depth' Turned off orgstruct++-mode (by M-x ... RET again) and filling of quoted text works fine as does normal writing. I will try (tomorrow) to isolate this with emacs -Q but my problem is that my gnus configuration is really really messy and difficult to isolate. Thanks, eric -- : Eric S Fraga (GnuPG: 0xC89193D8FFFCF67D) in Emacs 24.1.50.1 : using Org release_7.8.09-489-g541288
Re: [O] orgstuct++ does not lurk silently in the shadow
Eric Fraga e.fr...@ucl.ac.uk writes: unfortunately, I still have problems. I am using org completely up to date (a few minutes ago). me too, I just posted on the gnus.user list ... -- cheers, Thorsten
Re: [O] orgstuct++ does not lurk silently in the shadow
Hi Eric, Eric Fraga e.fr...@ucl.ac.uk writes: unfortunately, I still have problems. I am using org completely up to date (a few minutes ago). I do not turn orgstruct++-mode on automatically at all any more but instead turn it on manually by M-x orgstruct++-mode RET once I'm in a Message buffer. I tried it just now, replying to your email as my test. Turned on the mode, tried to fill the quoted text and it all got mixed up. You are using the exact same version of Gnus and Emacs than I do. Unless you forgot to reload the correct (uncompiled?) version of Org -- I barely dare suggesting this :) -- then this comes from something in your configuration. But no matter where it comes from, we should fix it! Undid that. Then went to the bottom and tried to insert a new line and got: org-indent-line-function: Lisp nesting exceeds `max-lisp-eval-depth' If you're certain you're loading Org correctly, please try the attached patch and just tell me if the error disappears. One possibility I can think of is that perhaps you turn on both orgstruct-mode and orgstruct++-mode. In that case when orgstruct++-mode will save the fill*/indent* variables from the _previous_ mode, it will suppose they are the ones from message-mode, which will not be true and will perhaps cause a loop. The attached patch prevents such a loop. Turned off orgstruct++-mode (by M-x ... RET again) and filling of quoted text works fine as does normal writing. This is already a progress against orgstruct++-mode as it worked before ... somewhat recomforting. I will try (tomorrow) to isolate this with emacs -Q but my problem is that my gnus configuration is really really messy and difficult to isolate. We just need this: C-x C-f ~/.gnus.el M-x occur RET message*hook Also check in your Emacs customization file. HTH, diff --git a/lisp/org.el b/lisp/org.el index e7c42db..17a5bc2 100644 --- a/lisp/org.el +++ b/lisp/org.el @@ -20534,7 +20534,8 @@ If point is in an inline task, mark that task instead. (org-move-to-column column) (when (and orgstruct-is-++ (eq pos (point))) (org-let org-fb-vars - '(indent-according-to-mode) + '(or (eq indent-line-function 'org-indent-line-function) + (indent-according-to-mode)) (defun org-indent-drawer () Indent the drawer at point. -- Bastien
Re: [O] orgstuct++ does not lurk silently in the shadow
Bastien b...@gnu.org writes: Hi Eric, Hello Bastien! [...] You are using the exact same version of Gnus and Emacs than I do. Unless you forgot to reload the correct (uncompiled?) version of Org -- I barely dare suggesting this :) -- then this comes from something in your configuration. Definitely using the latest org! And don't worry about asking: I have done sillier things in the past :-( Undid that. Then went to the bottom and tried to insert a new line and got: org-indent-line-function: Lisp nesting exceeds `max-lisp-eval-depth' If you're certain you're loading Org correctly, please try the attached patch and just tell me if the error disappears. The patch works! I have manually make orgstruct++-mode active: filling of paragraphs, including quoted ones, seems to work and there is no longer any infinite recursion. [...] We just need this: C-x C-f ~/.gnus.el M-x occur RET message*hook For completeness, I have no version of orgstruct++-mode or orgstruct-mode automatically turned on by any hook! Thanks again, eric -- : Eric S Fraga (GnuPG: 0xC89193D8FFFCF67D) in Emacs 24.1.50.1 : using Org release_7.8.09-529-g035ab3.dirty
Re: [O] orgstuct++ does not lurk silently in the shadow
Hi Nicolas, Nicolas Goaziou n.goaz...@gmail.com writes: Further, with orgstuct++-mode, trying to add a new line (i.e. my response), I get the following error message: , | org-indent-line-function: Lisp nesting exceeds `max-lisp-eval-depth' [3 times] ` I confirm the problem with latest Org (Org-mode version 7.8.09 (release_7.8.09-483-g01d83d)) and GNU Emacs 24.0.95.1. I pushed a fix -- can you test and confirm? Thanks, -- Bastien
Re: [O] orgstuct++ does not lurk silently in the shadow
Hello, Bastien b...@gnu.org writes: I pushed a fix -- can you test and confirm? I get the following error every time I try to indent: org-between-regexps-p: Variable binding depth exceeds max-specpdl-size Regards, -- Nicolas Goaziou
Re: [O] orgstuct++ does not lurk silently in the shadow
Hi Nicolas, Nicolas Goaziou n.goaz...@gmail.com writes: I pushed a fix -- can you test and confirm? I get the following error every time I try to indent: In org-mode or in message-mode? org-between-regexps-p: Variable binding depth exceeds max-specpdl-size Does (setq max-specpdl-size 1) help in any way? -- Bastien
Re: [O] orgstuct++ does not lurk silently in the shadow
Nicolas Goaziou n.goaz...@gmail.com writes: Bastien b...@gnu.org writes: I pushed a fix -- can you test and confirm? I get the following error every time I try to indent: org-between-regexps-p: Variable binding depth exceeds max-specpdl-size If you are in message-mode, please also provide the value of `org-fb-vars' and try to debug `org-indent-line-function' to make sure it stops at the first call of `org-between-regexps-p' (not at the ones that may be called inside the `orgstruct++-with-major-mode-filling' macro...) Thanks! -- Bastien
Re: [O] orgstuct++ does not lurk silently in the shadow
Bastien b...@gnu.org writes: Nicolas Goaziou n.goaz...@gmail.com writes: I get the following error every time I try to indent: In org-mode or in message-mode? I meant in Message mode. org-between-regexps-p: Variable binding depth exceeds max-specpdl-size Does (setq max-specpdl-size 1) help in any way? Not at all. After increasing `max-specpdl-size', I have to increase `max-lisp-eval-depth' and it still fails. I just have to wait longer. Regards, -- Nicolas Goaziou
Re: [O] orgstuct++ does not lurk silently in the shadow
Bastien b...@gnu.org writes: Nicolas Goaziou n.goaz...@gmail.com writes: Bastien b...@gnu.org writes: I pushed a fix -- can you test and confirm? I get the following error every time I try to indent: org-between-regexps-p: Variable binding depth exceeds max-specpdl-size If you are in message-mode, please also provide the value of `org-fb-vars' ((adaptive-fill-function (quote nil)) (adaptive-fill-regexp (quote \(\([ ]*[_.[:word:]]++\|[]*[]|]\)+\)[ ]*\|[ ]*\([-–!|#%;*·•‣⁃◦]+[ ]*\)*)) (fill-paragraph-function (quote message-fill-paragraph)) (paragraph-separate (quote --text follows this line--$\|[ ]*$\|-- $\|---+$\|^ $\|.*wrote:$\|\(\([ ]*[_.[:word:]]++\|[]*[]|]\)+\)[ ]*$\|#!*/?\(multipart\|part\|external\|mml\|secure\))) (auto-fill-inhibit-regexp (quote [ ]*\(|\|#\+\(tblfm\|orgtbl\|tblname\):\))) (paragraph-start (quote --text follows this line--$\|[]*$\|-- $\|---+$\|^$\|.*wrote:$\|\(\([ ]*[_.[:word:]]++\|[]*[]|]\)+\)[ ]*$\|#!*/?\(multipart\|part\|external\|mml\|secure\))) (indent-line-function (quote indent-relative)) (auto-fill-function (quote message-do-auto-fill))) and try to debug `org-indent-line-function' to make sure it stops at the first call of `org-between-regexps-p' (not at the ones that may be called inside the `orgstruct++-with-major-mode-filling' macro...) It's the same as the previous bug: when I try to debug `org-indent-line-function', the problem disappears. Regards, -- Nicolas Goaziou
Re: [O] orgstuct++ does not lurk silently in the shadow
Bastien b...@gnu.org writes: Hi Christopher, I've now fixed orgstuct++-mode so that you can turn it on and off safely. Some Org filling functions fall back on the major-mode filling functions through a new defmacro. Please test and report any problem. Thanks! Ummm, doesn't work well for me at all. If I try to fill your main paragraph above, the whole quoted text is filled into one mess. Without orgstuct++-mode, it works. Further, with orgstuct++-mode, trying to add a new line (i.e. my response), I get the following error message: , | org-indent-line-function: Lisp nesting exceeds `max-lisp-eval-depth' [3 times] ` with the following backtrace: , | Debugger entered--Lisp error: (error Lisp nesting exceeds `max-lisp-eval-depth') | org-indent-line-function() | org-indent-line-function() | org-indent-line-function() | ... repeated ad infinitum ... ` thanks, eric -- : Eric S Fraga (GnuPG: 0xC89193D8FFFCF67D) in Emacs 24.1.50.1 : using Org release_7.8.09-482-gf47915
Re: [O] orgstuct++ does not lurk silently in the shadow
Hi Eric, Eric Fraga e.fr...@ucl.ac.uk writes: Ummm, doesn't work well for me at all. If I try to fill your main paragraph above, the whole quoted text is filled into one mess. Without orgstuct++-mode, it works. Further, with orgstuct++-mode, trying to add a new line (i.e. my response), I get the following error message: , | org-indent-line-function: Lisp nesting exceeds `max-lisp-eval-depth' [3 times] ` with the following backtrace: , | Debugger entered--Lisp error: (error Lisp nesting exceeds `max-lisp-eval-depth') | org-indent-line-function() | org-indent-line-function() | org-indent-line-function() | ... repeated ad infinitum ... ` Mh.. do you confirm this is from latest git? More precisely, please make sure you have this commit in your git log: http://orgmode.org/w/?p=org-mode.git;a=commit;h=a8c312 I don't see how the Emacs version could interfere here, but please give it too -- along with the way you call orgstuct++-mode for messages. I do this: (add-hook 'message-mode-hook 'turn-on-orgstruct++) It works fine here. Best, -- Bastien
Re: [O] orgstuct++ does not lurk silently in the shadow
Hello, Eric Fraga e.fr...@ucl.ac.uk writes: Further, with orgstuct++-mode, trying to add a new line (i.e. my response), I get the following error message: , | org-indent-line-function: Lisp nesting exceeds `max-lisp-eval-depth' [3 times] ` I confirm the problem with latest Org (Org-mode version 7.8.09 (release_7.8.09-483-g01d83d)) and GNU Emacs 24.0.95.1. I also use orgstruct++-mode. Regards, -- Nicolas Goaziou
Re: [O] orgstuct++ does not lurk silently in the shadow
Hi Nicolas, Nicolas Goaziou n.goaz...@gmail.com writes: I confirm the problem with latest Org (Org-mode version 7.8.09 (release_7.8.09-483-g01d83d)) and GNU Emacs 24.0.95.1. I also use orgstruct++-mode. I can reproduce the problem with (add-hook 'message-setup-hook 'turn-on-orgstruct++) I cannot reproduce it with (add-hook 'message-mode-hook 'turn-on-orgstruct++) Can you confirm this is the same for you? (The manual advertizes using `message-mode-hook'.) Thanks, -- Bastien
Re: [O] orgstuct++ does not lurk silently in the shadow
Hello, Bastien b...@altern.org writes: Nicolas Goaziou n.goaz...@gmail.com writes: I can reproduce the problem with (add-hook 'message-setup-hook 'turn-on-orgstruct++) I cannot reproduce it with (add-hook 'message-mode-hook 'turn-on-orgstruct++) Can you confirm this is the same for you? No, it happens with `message-mode-hook'. Here is the relevant part of my configuration: #+begin_src elisp (defun ngz-message-mode-hook-handler () (turn-on-flyspell) (turn-on-orgtbl) (turn-on-orgstruct++) (set (make-local-variable 'org-footnote-tag-for-non-org-mode-files) nil) (local-set-key (kbd C-c C-x f) 'org-footnote-action)) (add-hook 'message-mode-hook 'ngz-message-mode-hook-handler) #+end_src Regards, -- Nicolas Goaziou
Re: [O] orgstuct++ does not lurk silently in the shadow
Nicolas Goaziou n.goaz...@gmail.com writes: No, it happens with `message-mode-hook'. Here is the relevant part of my configuration: I can't reproduce this with a bare emacs-24.0.94 -Q (this is the oldest Emacs I can test with.) Can you detail the steps to reproduce? Sorry this is boring, but I don't want to let such a serious but in the wild. Thanks! -- Bastien
Re: [O] orgstuct++ does not lurk silently in the shadow
Bastien b...@altern.org writes: Nicolas Goaziou n.goaz...@gmail.com writes: No, it happens with `message-mode-hook'. Here is the relevant part of my configuration: I can't reproduce this with a bare emacs-24.0.94 -Q (this is the oldest Emacs I can test with.) Can you detail the steps to reproduce? I cannot reproduce it with -Q (though indentation is strange at the beginning). I'm not sure where it comes from, though. It looks like an eisenbug to me. Debugging `org-indent-line-function' didn't help. Regards, -- Nicolas Goaziou
Re: [O] orgstuct++ does not lurk silently in the shadow
Hi Christopher, I've now fixed orgstuct++-mode so that you can turn it on and off safely. Some Org filling functions fall back on the major-mode filling functions through a new defmacro. Please test and report any problem. Thanks! -- Bastien
Re: [O] orgstuct++ does not lurk silently in the shadow
Bastien b...@gnu.org writes: Hi Eric, Eric Fraga e.fr...@ucl.ac.uk writes: Thanks for this. This behaviour has annoyed me for years now. Me too... I've never gotten around to figuring out why I got this behaviour because I get caught up with writing the email and forget by the time I finish... I hope somebody can indeed fix this as it sounds quite straightforward (for better elisp programmers than I...). This should be fixed now in master. Can you confirm? There is one remaining use-case I didn't fix: the case where you use M-q on the first line of the message (things are messed up then.) I don't know how to fix this, but this is a rather border-line case, as simply typing and using auto-fill does the right thing. Thanks for insisting on this! Bastian, Filling seems to work much better now but I don't know whether any of this is org or gnus. However, I am not sure what is happening here. With the following line in my configuration: (add-hook 'message-setup-hook 'turn-on-orgstruct++) all my emails end up indented strangely. Specifically, if this email had been composed with that hook entry, all my lines would have been indented 2 spaces. See my previous message to this list if you can find such easily... I commented out that line in my configuration and restarted Emacs to compose this reply. No indenting is happening (which is a good thing). Very strange. I do not understand why turning on orgstuct++ would cause a default indentation of all text. I am happy to explore this further if you give me some hint as to what to try. I am using gnus and org up to date as of yesterday (Australia time). I haven't updated today yet. Thanks, eric -- : Eric S Fraga (GnuPG: 0xC89193D8FFFCF67D) in Emacs 24.1.50.1 : using Org release_7.8.09-414-gb3e8a8
Re: [O] orgstuct++ does not lurk silently in the shadow
Eric Fraga e.fr...@ucl.ac.uk writes: [...] Bastian, Filling seems to work much better now but I don't know whether any of this is org or gnus. Actually, I just noticed from my modeline that OrgTbl is active. However, I have searched everything in my configuration and cannot see where I turn this on! I am perturbed :( The point being that orgstruct++-mode does cause problems with indentation and filling in email messages but without it, but with OrgTbl maybe, filling is working very nicely now. Thanks again, eric -- : Eric S Fraga (GnuPG: 0xC89193D8FFFCF67D) in Emacs 24.1.50.1 : using Org release_7.8.09-414-gb3e8a8
Re: [O] orgstuct++ does not lurk silently in the shadow
Hi Eric, Eric Fraga e.fr...@ucl.ac.uk writes: (add-hook 'message-setup-hook 'turn-on-orgstruct++) Try this instead, as advertized in the manual: (add-hook 'message-mode-hook 'turn-on-orgstruct++) HTH, -- Bastien
Re: [O] orgstuct++ does not lurk silently in the shadow
Bastien b...@gnu.org writes: This should be fixed now in master. Thank you so much for fixing this. There is still one major issue, though. Consider this text: , | rms | foo | | org-mode ` Place the point somewhere in the first or second line and do a M-q. Vanilla message-mode correctly recognises the paragraph prefix whereas an use of Org's enhancements breaks this. Other than that, disabling orgstruct++-mode after enabling it breaks everything. I get tons of Lisp nesting exceeds `max-lisp-eval-depth', even when doing simple things like moving around the buffer or trying to execute and extended command. For the record, I use Gnus and Org master (b5f4b52). This is how my integration is set up: #+begin_src emacs-lisp (add-hook 'gnus-message-setup-hook (lambda () (turn-on-orgtbl) (turn-on-orgstruct++) (set (make-local-variable 'org-footnote-auto-label) 'plain) (set (make-local-variable 'org-footnote-tag-for-non-org-mode-files) nil))) (define-key message-mode-map (kbd C-c f) (lambda () (interactive) (save-restriction (my-message-narrow-to-body) (org-footnote-action (add-hook 'message-send-hook (lambda () (save-restriction (my-message-narrow-to-body) (org-footnote-normalize #+end-src Christopher
Re: [O] orgstuct++ does not lurk silently in the shadow
Christopher Schmidt christop...@ch.ristopher.com writes: There is still one major issue, though. Consider this text: , | rms | foo | | org-mode ` Place the point somewhere in the first or second line and do a M-q. Vanilla message-mode correctly recognises the paragraph prefix whereas an use of Org's enhancements breaks this. It all boils down to orgstruct++-mode setting non-org-vars like paragraph-start or paragraph-separate to the org values. This is fine, even the doc string says so. ,[ C-h f orgstruct++-mode RET ] | Toggle `orgstruct-mode', the enhanced version of it. | In addition to setting orgstruct-mode, this also exports all indentation | and autofilling variables from org-mode into the buffer. It will also | recognize item context in multiline items. ` This does not work with message-mode and any other mode that imposes a special document structure, though. I think the docstring and the manual should note that explicitly, and the line promoting the use of turn-on-orgstruct++ in message-mode-hook in (info (org)Orgstruct mode) should be removed. Christopher
Re: [O] orgstuct++ does not lurk silently in the shadow
Hi Eric, Eric Fraga e.fr...@ucl.ac.uk writes: Thanks for this. This behaviour has annoyed me for years now. Me too... I've never gotten around to figuring out why I got this behaviour because I get caught up with writing the email and forget by the time I finish... I hope somebody can indeed fix this as it sounds quite straightforward (for better elisp programmers than I...). This should be fixed now in master. Can you confirm? There is one remaining use-case I didn't fix: the case where you use M-q on the first line of the message (things are messed up then.) I don't know how to fix this, but this is a rather border-line case, as simply typing and using auto-fill does the right thing. Thanks for insisting on this! -- Bastien
Re: [O] orgstuct++ does not lurk silently in the shadow
Christopher Schmidt christop...@ch.ristopher.com writes: Hi gurus, I have a problem with Org-mode version 7.8.03 (release_7.8.03.351.g47eb3) on GNU Emacs 24.0.94.1 (x86_64-unknown-linux-gnu, GTK+ Version 2.24.9) of 2012-03-08. (info (org)Orgstruct mode) says that one is allowed to use orgtbl-mode in message-mode. Unfortunately this does not work as expected. Recipe: emacs -q C-x m M-q # fill-paragraph ignores the message header eval: (add-hook 'message-mode-hook 'turn-on-orgstruct++) C-x m M-q # fill-paragraph does ignore the message header The bug is caused by fill-paragraph-function not being set back to message-fill-paragraph when orgstruct-hijacker-command-22 falls back to fill-paragraph. Christopher Thanks for this. This behaviour has annoyed me for years now. I've never gotten around to figuring out why I got this behaviour because I get caught up with writing the email and forget by the time I finish... I hope somebody can indeed fix this as it sounds quite straightforward (for better elisp programmers than I...). -- : Eric S Fraga (GnuPG: 0xC89193D8FFFCF67D) in Emacs 24.1.50.1 : using Org-mode version 7.8.06 (release_7.8.06.181.g67694.dirty)
[O] orgstuct++ does not lurk silently in the shadow
Hi gurus, I have a problem with Org-mode version 7.8.03 (release_7.8.03.351.g47eb3) on GNU Emacs 24.0.94.1 (x86_64-unknown-linux-gnu, GTK+ Version 2.24.9) of 2012-03-08. (info (org)Orgstruct mode) says that one is allowed to use orgtbl-mode in message-mode. Unfortunately this does not work as expected. Recipe: emacs -q C-x m M-q # fill-paragraph ignores the message header eval: (add-hook 'message-mode-hook 'turn-on-orgstruct++) C-x m M-q # fill-paragraph does ignore the message header The bug is caused by fill-paragraph-function not being set back to message-fill-paragraph when orgstruct-hijacker-command-22 falls back to fill-paragraph. Christopher