#foswiki 2017-01-19,Thu

↑back Search ←Prev date Next date→ Show only urls(Click on time to select a line by its url)

WhoWhatWhen
GithubBot[distro] gac410 created Item14281 (+1 new commit): https://git.io/vMMog
distro/Item14281 943beae George Clark: Item14281: Implement more complete control over cookies...
[03:53]
***GithubBot has left [03:53]
FoswikiBothttps://foswiki.org/Tasks/Item14281 [03:53]
............................................. (idle for 3h42mn)
***ChanServ sets mode: +o cdot [07:35]
ChanServ sets mode: +o MichaelDaum [07:43]
................................ (idle for 2h38mn)
foswiki_irc3IS MarkDown supported at all? [10:21]
jastno [10:23]
............ (idle for 55mn)
GithubBot[SocialSharePlugin] MichaelDaum pushed 1 new commit to master: https://git.io/vMD3y
SocialSharePlugin/master f831b07 MichaelDaum: Item14278: docu and final enhancements
[11:18]
***GithubBot has left [11:18]
FoswikiBothttps://foswiki.org/Tasks/Item14278 [ Item14278: implement a widget for social share buttons ] [11:18]
GithubBot[SocialSharePlugin] MichaelDaum pushed 1 new commit to master: https://git.io/vMD3A
SocialSharePlugin/master c5b6803 MichaelDaum: Item14278: prep rel
[11:19]
***GithubBot has left [11:19]
...... (idle for 25mn)
sahilsinha has quit IRC (Ping timeout: 256 seconds) [11:44]
............................ (idle for 2h17mn)
ChanServ sets mode: +o gac410 [14:01]
...... (idle for 28mn)
zak256In Support.Foswiki2MigrationIssues it says "Foswiki 2.0 bulk_copy.pl utility is broken!" Is this still the case? [14:29]
***ChanServ sets mode: +o Lynnwood [14:31]
gac410Okay, a lot has changed and I forgot that that topic existed. (One of the problems with wiki's ) [14:32]
zak256So, it's obsolete then? That's fine then :) [14:33]
gac410bulk_copy has generally been fixed, but it can be very slow. We do still recommend using CharsetConverter for really big sites. It's what we used on Foswiki.org [14:33]
zak256Well, it's only executed once, resp. some times for testing purposes. In the end I do not care if it takes some time. [14:34]
gac410bulk_copy depends upon the stored data really being 100% correct & perfect in structure. It checks out and copies each revision of each topic, going back to rev 1.
So if you have 1000's of topics with 1000's of revisions, it can take many many hours
[14:34]
zak256Yes, I saw that. On the first run I only converted the latest revision. [14:35]
gac410It also cannot copy anything that is "hidden" from the store API. Attachments manually put in pub subdirectories, attachments with leading _underscore, etc.
So it's a "very strict" copy, following all the restrictions of the store/API.
[14:35]
zak256I will keep that in mind, thanks. But generally it's no problem if not everything is perfect. The old wiki will stay some time, and if something went wrong that can be corrected manually afterwards. [14:36]
gac410CharsetConverter has some advantages. can do 1 web at a time. If you forgot to copy a web, you can go back and get it later.
And I need to run ... Appointment in a few minutes. Back in an hour or so.
[14:37]
zak256Bye [14:38]
gac410Oh... one little gotcha on plainfile store. If you "touch" files on disk that corrupts history. PlainFile store uses the file system's timestamps as part of the history rather than timestamps in the metadata.
Faster for sure! But sometimes one forgets ...
[14:39]
zak256I remember that has been discussed before. Isn't the timestamp stored within the files as well?
But this can be continued later. ;)
[14:40]
jastit is, usually, in the %META:TOPICINFO line
in fact in Modell Aachen's fork of PlainFileStoreContrib there's a helper script to fix the file timestamps based on TOPICINFO data
unfortunately I think it doesn't work with the official version of PFS
[14:42]
gac410Oh... if you do have topics with incorrect encodings, CharsetConverter may be useful even if you end up using the bulk_copy, as it can quickly inspect the files for encoding issues. [14:44]
Lynnwoodgac410 "Oh... one little gotcha on plainfile store. If you "touch" files on disk that corrupts history. " Does this mean that we can't ever directly modify topics on file level? [14:45]
gac410zak256: You also may want to convert using CP-1252 as the "from" encoding. That's pretty common if you have windows users cutting/pasting in smart quotes, etc.
Lynnwood: I'm not really sure of the extent of the issue. I just recall that there are some dragons lurking there.
[14:46]
Lynnwoodyikes! [14:47]
gac410PFS though does use the file system meta for the trusted information rather than the embedded meta. [14:47]
zak256Thanks for all the hints, I noted this.
It will become very relevant at a much later stage.
[14:48]
jastLynnwood: it's okay to directly modify the topics (same caveats as with RCS-based store, that is), just not the old revisions in the ,pfv subdir [14:48]
Lynnwoodah. ok [14:49]
jastnot the new revisions, either, I should say ;) [14:49]
zak256But wouldn't it be good if PFS would use the metadata timestamp instead of the filename timestamp then? [14:49]
Lynnwoodthat's not so bad. one was never able to edit earlier versions anyways and I can't see the need so much for that. [14:49]
jasttechnically using the file timestamp is faster, which is maybe why the author decided to do it that way [14:49]
zak256Ah, okay. [14:49]
jastI don't know about all the design decisions though
I'd see about porting the fix script to standard PFS but I don't really have any time to spare at all right now
[14:49]
zak256Where can I get that fix script for the forked version? [14:51]
jasthttps://github.com/modell-aachen/PlainFileStoreContrib/blob/release/riga/tools/touch_files.pl [14:55]
zak256Great thanks. If I manage to adapt this, I will share.
But it will be rather late this year I suppose.
[14:56]
jastthe reason it doesn't work without the fork is that most of the logic is in the PFS module itself
fixing that more or less means copying the code over into the script and rewriting it a little so all the references work
[14:57]
zak256You mean in the contributed module? Argh...
I will see...
[14:58]
jasthere: https://github.com/modell-aachen/PlainFileStoreContrib/blob/release/riga/lib/Foswiki/Store/PlainFile.pm#L1678 [14:58]
zak256What is the effect on the wiki, if the mtime changed on file system level? [15:09]
jaston a revision file? the revision will look as if it was created at that time [15:13]
zak256But no functional error?
I guess this will be acceptable then. As backup solution the script will be fine (in the future...) but if there will be no critical issues like problems accessing history or current files this is okay.
[15:16]
Lynnwoodgac410 - I'm going to give ImmediateNotifyPlugin another look today to see if I can determine where it is failing.
I turned on debug and it happily logs that it's working fine.
...just no notifications go out. :-)
[15:17]
................................. (idle for 2h40mn)
***zak256 has left [17:58]
..................... (idle for 1h40mn)
ChanServ sets mode: +o Lynnwood__ [19:38]
................ (idle for 1h18mn)
GithubBot[distro] gac410 pushed 1 new commit to master: https://git.io/vMyiI
distro/master 4009a9a George Clark: Item14279: Merge branch 'Release02x01'
[20:56]
***GithubBot has left [20:56]
FoswikiBothttps://foswiki.org/Tasks/Item14279 [ Item14279: CLI =tools/configure= fails to encode Password when run in prompting mode ] [20:56]

↑back Search ←Prev date Next date→ Show only urls(Click on time to select a line by its url)