This adds the whole new section in `org-roam-mode.el` for node dedicated
buffers, with a new global command `org-roam-buffer-display-dedicated`,
which supersedes `org-roam-buffer` command.
Along the way this also refactors the code for the persistent buffers to
clean it up, make it more clear in what it does, improve its robustness,
and to make it play nice with the dedicated buffers.
1. If it the link is at a headline, it shows the headline's content
instead of just repeating the headline
2. If the link is a bullet item in the plain-list, show the full list instead
org-roam-file-p throws `wrong-type-argument stringp nil` on buffers that
do not have a corresponding file. This fix changes org-roam-file-p to
return nil instead of crashing in this scenario.
Addresses #1645
Non-ASCII characters are displayed incorrectly because
`org-roam-unlinked-references-preview-line` uses
`insert-file-contents-literally` which ignores the encoding.
Use `insert-file-contents` which does the right thing.
fixes#1575
* (doc): Add section for C Compiler
In the Post-Installation Tasks section, I suggest to add a sub-section for the C
compiler requirement for emacsql-sqlite. I have put an explanation for Windows
-- I have tried to make it as succinct as I can without losing some fine points
that I believe important for users to avoid confusion.
I have just removed MSYS2 and the PATH on my Windows machine and repeated the
process as I described it here to confirm that it works.
Since this new part is distinct from the existing paragraphs for SQLite, I also
suggest a sub-section heading for them -- I used "SQLite" as its title.
Please feel feel to change any part as you see fit.
It might be also useful if macOS users added their parts. I believe XCode is the
easiest way to install a C compiler (I believe it would be clang for recent
XCode versions) or perhaps brew. I don't use macOS so I will refrain from adding
this part..
I did not add .texi file to this commit -- last time, I believe you had to
revise the texi file that my machine generated anyway. Let me know if I should
follow up with a texi file.
* remove reference to emacsql-sqlite3
Co-authored-by: Jethro Kuan <jethrokuan95@gmail.com>
This mostly goes over docstrings, updating them in places where it was
outdated and adding them to places where they didn't exist or were
existed as placeholders.
This brings them more in line with how other Org keywords, such as
\#+PROPERTY, are declared.
Previously
#+roam_alias: abc def
#+roam_alias: ghi
would result in only the last one ("ghi") being extracted. Now ("abc"
"def" "ghi") are all extracted (in that order).
* org-roam.el (org-roam--extract-tags-prop, org-roam--extract-titles-alias): Accept and return all values in a list, not just from one line.
(org-roam--extract-prop-as-list): New function. List prop extraction refactored from `org-roam--extract-tags-prop` and `org-roam--extract-titles-alias`
* tests/test-org-roam.el: Add tests for defining tags and aliases in multiple lines
* (doc): Add file-truename to set org-roam-directory
Refer this debuging Slack exchange:
https://orgroam.slack.com/archives/CV160S8EL/p1619089118195300
Not using `file-truename` to set
`org-roam-directory` can lead to an issue that is
hard to identify.
It appears that cache database may updates with
files but `org-roam-buffer` fails to find the file
as symbolic links do not resolve. This is
confusing as the table query to the table seems to
return what appears to be correct entries, but the
backlink fails to insert context around the
link (as Org-roam fails to find the file).
There have been similiar issues -- by making sure
`file-truename` is added in the documentation and
README, it is hoped to eliminate such issues to
recur.
This is probably relevant for V2.
* undo changes to texi
Co-authored-by: Jethro Kuan <jethrokuan95@gmail.com>
* Normalize slug using ucs-normalize-NFC-string after remove Unicode spacing mark
* Add org-roam-slug--preserve-chars-from-normalization for Unicode Normalization
* Add org-roam-slug-trim-chars instead of org-roam-slug--preserve-chars-from-normalization
Instead of storing preview content in the database, now provide a
function to fetch these on the fly. This paves the way for future
improvements (e.g. showing more lines)
