We got off on a bit of a tangent here testing mw-render, but it seems to work. As a result, tho, the log is a little longer than it should be.
00:20:03 <jjmcd> #startmeeting Converting wiki to git 00:20:04 <zodbot> Meeting started Thu Mar 18 00:20:03 2010 UTC. The chair is jjmcd. Information about MeetBot at http://wiki.debian.org/MeetBot. 00:20:05 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic. 00:20:15 <jjmcd> #chair stickster, laubersm 00:20:15 <zodbot> Current chairs: jjmcd laubersm stickster 00:20:40 * jjmcd is typing url on other machine 00:20:47 <laubersm> ok - jjmcd correct me if I am wrong here... first I want to make sure I have a current git pull of r-n and am on the correct branch... 00:20:58 <jjmcd> yeah 00:20:59 <laubersm> Then I want to open wiki to the correct page... 00:21:12 <jjmcd> To begin, you need to have git and publican installed 00:21:14 <laubersm> care to review for the crowd the details on those and what == correct 00:21:20 <jjmcd> yep 00:21:34 <jjmcd> To get the release notes 00:21:58 <jjmcd> git clone ssh://git.fedorahosted.org/git/docs/release-notes.git 00:22:15 <jjmcd> If you don't have commit access, use git: instead of ssh: 00:22:19 * laubersm notes cleaning out subdirs from F12 test builds is also a good idea - thanks to git status 00:22:49 <jjmcd> The git clone will make a release-notes directory 00:23:06 <laubersm> er... it was f11 test builds... I *knew* I wouldn't be starting on the correct branch :) 00:23:34 <jjmcd> and everything is in there, so if you do this from ~/Documents/Release-Notes, then the release notes will be in ~/Documents/Release-Notes/release-notes 00:23:45 <stickster> jjmcd: There's a bunch of stuff in build/ that may not belong anymore if I'm correct 00:24:19 <stickster> not sure now. 00:24:29 <jjmcd> Some is still useful, some not 00:24:43 <jjmcd> We can get rid of the whole merging thing 00:24:54 * laubersm notes for those with an old version checked out - git status ; git branch -r ; git checkout master 00:26:24 * jsmith checks out the latest version 00:26:27 <jsmith> (slowly) 00:26:37 <jjmcd> How bad is this python thing going to break my system? 00:27:50 <jjmcd> The xml files will be in release-notes/en-US 00:28:12 <jjmcd> There is pretty much one file per beat, and they are named with pretty obvious names 00:28:48 <jjmcd> Things like MailServer.xml 00:30:19 <jjmcd> Taking the git business first, you edit the file, build the documents, check for errors, when you are satisfied... 00:30:33 <jjmcd> git status should show you only one file needs to be committed 00:30:48 <jjmcd> If that is the case, git add -u will add it to the commit list 00:30:52 <stickster> Looks like mw-render is working 00:31:31 <jjmcd> Keeps telling me I shouldn't trust this guy, but it is taking it's sweet time to install 00:31:32 <stickster> #info command line for mw-render: mw-render -c http://fedoraproject.org/w/ -o <outputfile> -w docbook <Wiki_Article_Name> 00:31:48 <jjmcd> You can add several files if you wish 00:32:08 <jjmcd> But you would like to commit a related group that you can describe as one thing. 00:32:22 <laubersm> content is coming from the wiki 00:32:25 <laubersm> #link http://fedoraproject.org/wiki/Documentation_Beats 00:32:33 <jjmcd> So, to do the commit, git commit -m "I fixed this and that" 00:32:36 <stickster> rudi: ^^^ above re: mw-render which may help get beats into XML faster 00:32:44 <jjmcd> Yes, I was going to fo there after git 00:32:49 <stickster> You have to massage the XML afterward but it's easier than doing individual copy/pastes 00:33:47 <rudi> Ta stickster 00:34:18 <jjmcd> still installing. It complained a lot about not trusting that guy! 00:38:11 <quaid> +1 on meetbot logging + #command usage = email with details to list so others are in the know 00:38:27 <jjmcd> Yuck 00:38:41 <jjmcd> Every {{package gets an inlinemediaobject 00:39:16 <jjmcd> stickster, I'm not so sure that is easier 00:39:38 <stickster> jjmcd: Those oughtta be removed 00:39:50 <stickster> We used to make it clear that release notes beats didn't get written like any other page 00:39:54 <stickster> i.e. less markup 00:40:00 <jjmcd> Yeah, they really mung up the works 00:40:16 <stickster> jjmcd: I'd suggest edit the wiki first, remove extraneous markup, save, then import. 00:41:18 <stickster> jjmcd: If it were possible, we could add a template to every page that would include that guidance -- the problem would be, I think that when you ran mw-render, that would appear in the output. 00:44:09 <quaid> #link https://fedoraproject.org/wiki/Converting_wiki_to_DocBook_XML#Converting_fro... 00:44:18 <quaid> in particular 00:44:20 <quaid> #link https://fedoraproject.org/wiki/Converting_wiki_to_DocBook_XML#Converting_to_... 00:44:25 <quaid> I just updated them for mw-render 00:44:35 <quaid> with steps for a quick for-loop to render a pile of pages, etc. 00:45:00 * quaid cleans up more cruft from that page 00:45:40 <stickster> jjmcd: Want help cleaning up those wiki pages for markup? I can start at the bottom of the release notes list and work my way up. 00:45:51 * stickster looking at https://fedoraproject.org/wiki/Documentation_Beats for the list 00:46:04 <jjmcd> I'm just trying to get a simple one to work 00:46:26 <stickster> jjmcd: What are you trying? 00:46:33 <jjmcd> mail 00:46:46 <stickster> Paste your command line 00:47:18 <jjmcd> It converts, but the result won't build 00:47:30 <stickster> jjmcd: Paste the XML to fpaste.org 00:47:33 <jjmcd> and the new publican is even worse than the old with error messages 00:47:59 <stickster> jjmcd: or... 00:48:02 <stickster> We can use gobby. 00:48:08 <stickster> We have all these cool collaboration tools, let's use them. 00:48:14 <jjmcd> K8gu 00:48:31 <jjmcd> This is a different box than my publican 1.0 box 00:48:34 * stickster recommends real links 00:48:46 <jjmcd> so cutting and pasting is messy 00:48:49 <stickster> #link http://fpaste.org/K8gu/ 00:48:55 <jjmcd> This box is way slow 00:49:30 <jjmcd> The xml looks fine but publican doesn't care for it 00:49:31 <stickster> jjmcd: Do you need a faster one? 00:49:47 <jjmcd> Just flipping between boxes 00:49:58 <jjmcd> This box is slow but has lots of screen, F11 00:50:00 <jsmith> I can give you a machine in the cloud, if you need one 00:50:13 <jjmcd> I pretty much do IRC and gwibber on it 00:50:24 <jjmcd> I do most stuff on my laptop 00:50:45 <quaid> jjmcd: the resulting XML probably won't build in publican as-is 00:50:50 <jjmcd> but I have a fast box with Publican 0.44, waiting till now to move it to f12 00:50:52 <quaid> it might build with raw docbook on the CLI though 00:51:13 <jjmcd> Well, that is my issue, but I really don't see anything wrong 00:51:34 <quaid> <sectioninfo> is new 00:51:43 <jjmcd> and wrong? 00:51:54 <quaid> not likely 00:51:56 <stickster> quaid: jjmcd: The DTD is wrong at top, that might not help. 00:52:35 <jjmcd> Oh, it had all that beat crap, I just cut and pasted the middle 00:52:41 <jjmcd> AHHHH 00:52:49 <jjmcd> builds if you get rid of sectioninfo 00:52:49 <quaid> does the original page have all the content in == sections for each piece == ? 00:53:44 <quaid> jjmcd: what tool was breaking on it? 00:53:47 <stickster> jjmcd: What should I be doing with beats that have tables of package changes? ship those changes into some other "all changes" page? 00:54:02 <jjmcd> stickster - those still need editing 00:54:05 <quaid> I bet <sectioninfo> is not on Publican's allowed tags list 00:54:16 <jjmcd> those tables only provide a clue for the beat writer 00:54:18 <stickster> quaid: Yeah, likely not 00:54:38 <stickster> s/<?/sectioninfo>//g 00:54:41 <jjmcd> OK, it probably only takes twice as long as cut/paste 00:55:53 <jjmcd> It also doesn't put the type= into the ulinks, but publican seems ok with that 00:56:27 <stickster> Wow, a blank kernel beat. That's not kool. 00:56:51 <quaid> we been using devel-announced? 00:57:03 <stickster> yup 00:57:15 <quaid> got a beat writer to ritually flog with a wet noodle? 00:57:18 <stickster> http://lists.fedoraproject.org/pipermail/devel-announce/2010-March/000589.ht... 00:57:22 <stickster> quaid: nope. 00:57:27 <jjmcd> stickster, use the status page 00:57:30 <quaid> awp 00:57:53 * quaid makes UNIX caveman sounds, 'AWK AWK GREP SED GREP' 00:58:01 <jjmcd> https://fedoraproject.org/wiki/Documentation_Beat_status 00:58:06 <quaid> thx 00:58:12 <stickster> No one listed for kernel. 00:59:13 <quaid> we could use to make some developer contacts in that list, people to reach out to individually when we see important holes 00:59:33 <stickster> "use to make"? :-D 00:59:45 <quaid> colloquialland 00:59:54 * stickster no longer lives there :-\ 01:00:00 <quaid> I think I'm in New Jersey tonight 01:00:23 <jjmcd> The package database has been helpful for me. The maintainers have been good with answers, not so good about providing content 01:00:45 <quaid> the maintainers need to help find the beat writer :) 01:00:51 <jjmcd> The all changes page has links to the package database, but it still takes a few clicks to get to the maintainers email 01:01:24 <quaid> there is someone in their extended circle ... if Dave Jones or Uli Drepper asked folks to write stuff for them, you think someone might o it. 01:02:33 <quaid> jjmcd: so, what I'm thinking is putting a named developer contact for each beat; it could be 'anaconda-devel' or 'Dave Jones' 01:02:42 <quaid> and then we email them :) 01:03:01 <jjmcd> There is a column for that, but it is mostly empty and not very helpful 01:03:11 <jjmcd> So far, the packagedb has been better 01:03:25 * laubersm notes we are off the topic of converting wiki to xml and documenting said process for new folks to be able to help out. 01:03:28 <jjmcd> Because it tells you someone currently working on it 01:03:38 <jjmcd> laubersm, yes 01:04:02 <jjmcd> OK, so you run mw-render 01:04:14 <jjmcd> Whoops, first un-tag the beat 01:04:19 <jjmcd> then run mw-render 01:04:36 <jjmcd> then cut and paste the "body" of the result into the xml 01:04:38 <laubersm> explain un-tag the beat please 01:04:53 <jjmcd> Remove {{package| macros 01:05:11 * laubersm assumes someone may also be new to wiki edits 01:05:22 <laubersm> thanks 01:05:31 <jjmcd> OK, back to the xml 01:05:40 <jjmcd> remove sectioninfo 01:05:54 <jjmcd> put back in <package> tags 01:06:01 <jjmcd> if all goes well 01:06:16 <jjmcd> publican --langs=en-US --formats=html should run 01:06:26 <jjmcd> look at result 01:06:32 <jjmcd> If you like it 01:06:37 <jjmcd> git add -u 01:06:46 <jjmcd> git commit -m "what I did" 01:06:50 <jjmcd> git push origin 01:07:05 <jjmcd> Pretty straightforward, actually 01:07:25 <jjmcd> So far, most of the beats are simple 01:07:43 <stickster> laubersm: Replacing {{package|foo}} with ''foo'' can help. Then it's easy to find the text "foo" because it will have <emphasis> tags around it you can just change to <package> 01:07:47 <jjmcd> It may get interesting if some of the beats have lists or tables or whatever 01:07:55 <jjmcd> Good catch 01:08:07 <laubersm> ok, lets walk through the Printing beat as an example 01:08:08 * stickster recalls that's why we avoided them previously 01:08:12 <stickster> So we could easily port to XML 01:08:31 <jjmcd> Yeah, even cutting and pasting is a little messy 01:08:36 * stickster edits the English. 01:08:55 <laubersm> No package tags so that should be real easy. I assume we can leave the admon since we aer just ignoring that part of the render 01:09:15 <jjmcd> OK, in this case, Emad78 left the link to the feature page 01:09:33 <jjmcd> We may want to remove it, or refer to it in the text 01:09:46 * stickster fixes the English 01:09:49 <jjmcd> We also want to get rid of the admonition 01:10:25 <jjmcd> Looks pretty good 01:10:34 <laubersm> I assume if we refernce the feature page we want it with the full external link? 01:10:40 <jjmcd> correct 01:10:47 <stickster> Yes 01:10:49 <jjmcd> Notice stickster caught that 01:11:07 <stickster> #info when editing, replace wiki links with full https:// links. 01:11:31 <jjmcd> #info also, remove admonition, and make sure there is an appropriate title 01:11:38 <laubersm> https or http? 01:11:50 <stickster> oh good call 01:11:56 <stickster> http will prevent overload on release day. 01:12:04 * stickster leaves that edit to laubersm 01:12:18 <stickster> #info when removing {{package|foo}}, it's helpful to replace it with ''foo'' (wiki italics) so it's easier to locate by the resulting <emphasis> tags in the XML 01:12:32 <jjmcd> #info prefer http rather than https for the printed release notes 01:14:25 <jjmcd> so the command is 01:15:11 <jjmcd> mw-render -c http://fedoraproject.org/w/ -o printtemp.xml -w docbook Documentation_Printing_Beat 01:15:24 <jjmcd> what did I miss? 01:16:53 <stickster> worksforme 01:17:29 * laubersm is back... ok - edits look good. trying the mw-render command now... 01:17:38 <stickster> jjmcd: You can do this 01:17:53 <stickster> mw-render -c http://fedoraproject.org/w/ -o printtemp.xml -w docbook -t Printing Documentation_Printing_Beat 01:18:01 <jjmcd> So, you copy all the text from the <section> before the first title, to the penultimate </section> 01:18:07 <stickster> sorry 01:18:15 <stickster> mw-render -c http://fedoraproject.org/w/ -o printtemp.xml -w docbook --title="Printing" Documentation_Printing_Beat 01:18:26 <jjmcd> and paste it into the middle of the en=US/Printing.xml 01:18:37 <stickster> Oh never mind, that sucks worse 01:18:43 <jjmcd> stickster, yes, I did MailServers and pushed it 01:18:53 <jjmcd> s/Beat/Beat" 01:19:32 <jjmcd> I think you can get away without the -t 01:19:47 <jjmcd> At least it worked for me 01:19:52 <stickster> Yeah, it adds a <booktitle> which is totally not wanted. 01:19:57 <stickster> Forget that option :-D 01:20:17 <jjmcd> Well, there is so much trash at the front anyway 01:20:56 <jjmcd> rudi has the xml all marked up with development stuff that will disappear when we get rid of status="draft" 01:21:17 <jjmcd> So we want to paste into rudi's files rather than replace them 01:21:24 <rudi> Yeah :) 01:22:10 <stickster> Yup yup 01:22:49 <jjmcd> laubersm, you may need to do a pull ... I pushed mailservers after you cloned 01:24:03 <laubersm> jjmcd, I got that.... 01:25:08 <laubersm> I'm multitasking and I really should just concentrate on this for a few minutes... lets see if I can build 01:25:15 <jjmcd> that darned peer anyway 01:26:49 <laubersm> for the record - the publican command you listed before is incomplete... 01:26:54 <jjmcd> ? 01:27:00 <jjmcd> worksforme 01:27:01 <laubersm> publican build --lang=en-US --format=html 01:27:06 <laubersm> forgot the word buid 01:27:10 <jjmcd> langs, formats 01:27:11 <laubersm> build :) 01:27:15 <jjmcd> ahhhhh 01:27:37 <laubersm> ok = one more time complete" 01:27:51 <jjmcd> #info publican build --langs=en-US --formats=html 01:27:54 <laubersm> publican build --langs=en-US --formats=html 01:28:12 * laubersm goes to fix markup error from attempt 01:29:04 <quaid> #link https://fedoraproject.org/wiki/User:Quaid/Documentation_Beat_status 01:29:20 <quaid> just looking to see if it was any more illuminating than https://fedoraproject.org/wiki/Documentation_Beats 01:29:47 <jjmcd> did you see Documentation_Beat_status? 01:30:02 <quaid> jjmcd: yeah, my table is drawn right from that 01:30:06 <jjmcd> Has pointers for newbies 01:30:12 <rudi_> Protip -- save a little typing: "publican build -l en-US -f html" 01:30:25 <quaid> I was trying to illuminate where people are needed specifically 01:30:30 <quaid> people or help 01:30:32 <jjmcd> Oh cool, both table and shorthand 01:30:47 * jjmcd was too lazy to format a table ;-) 01:31:08 <laubersm> #info when cut/paste - be sure that lines still wrap and do not have newlines inserted 01:31:17 <laubersm> arg 01:31:27 <jjmcd> must not be using emacs 01:32:46 <laubersm> nope - and I think I was demo-ing some vi set commands in class a couple of weeks ago and ended up on my machine instead of the demo box... 01:33:19 <jjmcd> First thing I do on a new install is delete vi 01:33:35 <jjmcd> and next symlink it to emacs 01:34:05 <laubersm> I'm sorry 01:34:11 <laubersm> pushed printing update 01:34:52 <jjmcd> OK, now go to the beat status page and mark it in xml 01:35:02 <jjmcd> and to the beats page and put a 1 in the xml column 01:35:20 <jjmcd> quaid, did you replace d_b_s with your table? 01:36:42 <jjmcd> laubersm, and it builds 01:36:53 <laubersm> jjmcd, of course 01:37:02 <laubersm> I even looked at the html page 01:37:14 <jjmcd> looks good 01:37:49 <jjmcd> admission: jjmcd is making html-single because it is easier when the doc is still small 01:38:06 <jjmcd> Not so good when it takes 10 minutes to load tho 01:38:37 <laubersm> ok - I got the 1 under xml column in beats page... 01:38:46 <jjmcd> rudi_, it is so much nicer to do ls en-US and not have to look at a hundred table- files 01:39:13 <rudi_> *nods :) 01:39:41 <jjmcd> laubersm, and the "in xml" on the status page? 01:39:43 <laubersm> and I got the status page 01:39:47 <jjmcd> ok, cool 01:40:24 <laubersm> that dual place to update in wiki is going away soon ??? (I hope) 01:40:31 <jjmcd> yes 01:40:46 <jjmcd> the dual thing is really only a way to get the new contributors started 01:41:00 <jjmcd> we have quite a few new and relatively new folks 01:41:21 <jjmcd> And I don't want then discouraged because of no way to contribute 01:42:11 <laubersm> ok... You guys should be done with all this moving to xml before I get up in the morning... but if not I can probably put in an hour or so tomorrow afternoon 01:42:20 <jjmcd> This way they can dive into anything that says opportunity,and if they are feeling their vinegar, possible opportunity 01:42:55 <jjmcd> OK, I can probably only do a couple more before bed, but there aren't that many ready 01:43:10 <jjmcd> Quite a few still need content 01:43:19 <laubersm> If they add content, I can help get it converted to docbook... or help them if they can push 01:43:49 <jjmcd> good deal 01:44:04 <jjmcd> Should we go ahead and do endmeeting, or wait a week or so? 01:44:11 <laubersm> lol 01:44:55 <stickster> heh 01:44:55 <laubersm> one more question... when does all this get into git by? And when is the next window of conversion? 01:45:06 <stickster> laubersm: yesterday :-D 01:45:22 <laubersm> oh yeah... nevermind 01:45:28 <jjmcd> We want to make POTs tomorrow night. We will make an RPM Wednesday. 01:45:34 <stickster> jjmcd: laubersm: I'd suggest that everyone involved have watches on all the beats 01:45:43 <stickster> So if someone changes the content we can do what we need to get the changes into POT 01:45:53 <jjmcd> I obviously do 01:46:11 <stickster> #info We shouldn't just copy new content over the old, because we might accidentally bring in trivial changes that aggravate translators 01:47:18 <quaid> jjmcd: nope, I didn't replace it with my table 01:47:55 * stickster looking for a beat, somebody hit me 01:47:55 <jjmcd> 'lo sizemj, you just missed our howto 01:48:07 <jjmcd> Sci/Tech, or you want an easy one? 01:48:29 <sizemj> I know was late 01:48:59 <jjmcd> I'll be posting the log in a couple of minutes. 01:49:05 <jjmcd> #endmeeting
docs@lists.stg.fedoraproject.org