diff options
Diffstat (limited to 'doc/forum')
4 files changed, 114 insertions, 0 deletions
diff --git a/doc/forum/Darcs_as_the_RCS___63__.mdwn b/doc/forum/Darcs_as_the_RCS___63__.mdwn index 2635690f7..9664240ee 100644 --- a/doc/forum/Darcs_as_the_RCS___63__.mdwn +++ b/doc/forum/Darcs_as_the_RCS___63__.mdwn @@ -9,3 +9,5 @@ What should I put in the configuration file to use darcs ? > Darcs is not yet supported. It's being [[worked_on|todo/darcs]]. > > That's good news for me then ! Thank you. + +>>> Better news: It will be in version 2.10. --[[Joey]] diff --git a/doc/forum/How_does_ikiwiki_remember_times__63__.mdwn b/doc/forum/How_does_ikiwiki_remember_times__63__.mdwn new file mode 100644 index 000000000..5522cbf45 --- /dev/null +++ b/doc/forum/How_does_ikiwiki_remember_times__63__.mdwn @@ -0,0 +1,89 @@ +This is similar to the last post in this forum. I want to know exactly how ikiwiki remembers the times associated with pages, especially when using it for blogging, so I know whether I can trust it or not. From that last thread, I think what ikiwiki does is this: + +* The created time of a file is when that file was first committed into the versioning repository (in my case git) + + > If `--getctime` it used, yes. In normal operation, when new files + > are added, ikiwiki sets the creation time to the ctime of the file + > on disk, rather than bothering to ask the VCS. --[[Joey]] + +* The modified time of a file is what that file was last updated in the repository + + > Almost right, the modified time is actually taken from the + > modification time of the file in disk. --[[Joey]] + +And with a blog, by default, the posts are ordered by creation time, although an option can order them by modified time. + +Okay. So this should mean that the times are safe if, for example, I delete my working copy and then clone another one from the bare git repository, or otherwise mess up the creation times and mtimes stored as file metadata on the filesystem. + +Do I have it right? + +> Some VCS, like git, set the file mtimes to the current time +> when making a new checkout, so they will be lost if you do that. +> The creation times can be retrived using the `--getctime` option. +> I suppose it might be nice if there were a `--getmtime` that pulled +> true modification times out of the VCS, but I haven't found it a big +> deal in practice for the last modification times to be updated to the +> current time when rebuilding a wiki like this. --[[Joey]] +> +> > Thanks for the clarification. I ran some tests of my own to make sure I understand it right, and I'm satisfied +> > that the order of posts in my blog can be retrieved from the VCS using the `--getctime` option, at least if I +> > choose to order my posts by creation time rather than modification time. But I now know that I can't rely on +> > page modification times in ikiwiki as these can be lost permanently. +> > +> > I would suggest that there should at least be a `--getmtime` option like you describe, and perhaps that +> > `--getctime` and `--getmtime` be _on by default_. In my opinion the creation times and modification times of +> > pages in ikiwiki are part of the user's content and are important to protect, because the user may be relying +> > on them, especially if they use blogging or lists of recently modified pages, etc. Right now the modification +> > times can be lost permanently. +> > +> > Is there a typo in the description of `--getctime` in the man page? +> > +> > > --getctime +> > > Pull **last changed time** for each new page out of the revision +> > > control system. This rarely used option provides a way to get +> > > the real creation times of items in weblogs, such as when build‐ +> > > ing a wiki from a new Subversion checkout. It is unoptimised and +> > > quite slow. It is best used with --rebuild, to force ikiwiki to +> > > get the ctime for all pages. +> > +> > Surely it is not the _last changed time_ but the _first seen time_ of each page that is pulled out of the VCS? +> > If the aim is to get the real creation times of items in weblogs, then the last times that the items were +> > changed in the VCS is not going to help. -- [[seanh]] +>>> Typo, fixed. --[[Joey]] + +> > > If you want to preserve the date of a page, the best way to do it is to +> > > use [[ikiwiki/directive/meta]] date="foo". This will survive checkouts, +> > > VCS migrations, etc. -- [[Jon]] +> > > +> > > > That's a good tip Jon. That would also survive renaming a page by renaming its mdwn file, which would +> > > > normally lose the times also. (And in that case I think both times are irretrievable, even by +> > > > `--getctime`). I might start using a simple script to make blog posts that creates a file for +> > > > me, puts today's date in the file as a meta, and opens the file in my editor. -- [[seanh]] + +>>>>> I use a script that does that and also sets up templates and tags +>>>>> for a new item: + + #!/bin/sh + set -u + set -e + + if [ $# -ne 1 ]; then + echo usage: $0 pagename >&2 + exit 1 + fi + + pagename="$1" + + if [ -e "$pagename" ]; then + echo error: "$pagename" exists >&2 + exit 1 + fi + + date=$(date) + echo '\[[!template id=draft]]' >> "$pagename" + echo "\[[!meta date=\"$date\"]]" >> "$pagename" + echo "\[[!tag draft]]" >> "$pagename" + git add "$pagename" + $EDITOR "$pagename" + +>>>>> -- [[Jon]] diff --git a/doc/forum/How_to_fix___34__does_not_map_to_Unicode__34___errors__63__.mdwn b/doc/forum/How_to_fix___34__does_not_map_to_Unicode__34___errors__63__.mdwn new file mode 100644 index 000000000..0b3895357 --- /dev/null +++ b/doc/forum/How_to_fix___34__does_not_map_to_Unicode__34___errors__63__.mdwn @@ -0,0 +1,20 @@ +I'm getting a number of errors like this when running ikiwiki: + + utf8 "\xA2" does not map to Unicode at /usr/local/share/perl/5.10.0/IkiWiki.pm line 739, <$in> chunk 1. + +I think it's because some of my files contain non-utf8, non-unicode, or somehow bad characters in them, probably fancy quotes and the like that have been copy-and-pasted from my web browser. The problem is that I have hundreds of files, I transferred them all over from pyblosxom to ikiwiki at once, and the error message doesn't tell me which file the error comes from. How can I fix this? + +Thanks +-- seanh + +> Unfortunatly, these messages are logged by perl so there's no way to add +> a filename to them. +> +> If you run the build in --verbose mode, you should see which page ikiwiki +> is working on, and unless it inlines some other page, you can be pretty +> sure that page contains invalid utf-8 if the message is then printed. +> +> Another option is to use the `isutf8` program from +> moreutils](http://kitenet.net/~joey/code/moreutils/), +> and run it on each file, it will tell you the line number +> and character position that is invalid. --[[Joey]] diff --git a/doc/forum/managing_todo_lists.mdwn b/doc/forum/managing_todo_lists.mdwn index b4bbac255..0a69af805 100644 --- a/doc/forum/managing_todo_lists.mdwn +++ b/doc/forum/managing_todo_lists.mdwn @@ -39,3 +39,6 @@ sure how to handle embeds or challenges from the CGI such as a login challenge >> thanks - I'll give it a look. I spent a few hours writing some javascript to manipulate a ul/li DOM tree in an outliner-fashion the other day. I might be able to join the puzzle pieces together sometime. [[Jon]] a solution for this could be similar to a solution for [[todo/structured page data]], as todo lists are definitely a form of structured data. (in both cases, the page's current content is rendered into a html form, whose result is then saved as the page's new contents) --[[chrysn]] + +> Thanks for the link: yup, there's definitely some common ground there. +> -- [[Jon]] |