I'll try to see what I can do with this, but it doesn't address my original request below.

We really need to meet in real time on this. I'll start another thread on that.

John

On 12/11/2009 02:55 PM, John J. McDonough wrote:

John

Looking at it with my Release Notes blinders on, I think we need to first focus on the major tasks, AND THEIR RELATIONSHIPS. I'm sort of ignoring docs.fp.o at this point because that happens almost automatically as a result of having the RPM.

Starting from the end ...

We need to have the RPM built BEFORE the RC Compose. Somehow we managed to miss this little detail for F12. There are always hiccups so I think we should leave 3 days for this. We *might* end up with another RPM for the "Bug Complete" compose, but since we will have addressed the issues already, we can probably get by with a day for that.

Backing up from the RC RPM, it looks as if we don't gain a lot by trying to get all the translation done for beta like we did for F12. The time for GA is about the same as the time for Beta. So let's assume we have 10 days for translation (which must be finished before we make the RPM), and another 10 days for post-beta edits, that brings us back about to the Beta compose, so basically, it is the maximum amount of time we have.

So the basics are:

• We need to have the RPM **before** RelEng's compose
• We need the translation before the RPM
• We need the text before we can translate it

The only difference between Beta and GA is that for Beta we probably should take an extra few days for unclaimed beats.

Alpha is kind of a minor event since we don't translate it, and in the past we haven't made an RPM (although I think we should).

Take a look at http://jjmcd.fedorapeople.org/F13-RelNotes-MajorTasks.html

there are lots of other things we need to do for Release Notes, but these are the things that essentially set the schedule. Unfortunately, the html export doesn't show the dependencies, which are the main thing. But all of the other tasks are mostly noise. It is this business of write, translate, make the RPM that sets the schedule, and we need to give Jesse the RPM before he composes, so RelEng's composes are major milestones as far as we are concerned.

Also, there is an assumption here that the new Transifex is in place. We have also had discussions about sending POTs much more frequently. The way we have been handling it L10N gets the MINIMUM amount of time to do the translations. Often we have plenty of changes backed up before we can reveal them to L10N. This doesn't seem like the best plan. We have an accounting problem trying to keep all the partial and un-done translations straight, and L10N is as rushed as possible.

--McD

On Fri, 2009-12-11 at 12:22 -0800, John Poelstra wrote:

...

In anticipation of the meeting suggested by Sparks on logistics list, I've created a text version of the schedule that should work well for customizing or using in gobby.

http://poelstra.fedorapeople.org/schedules/f-13-draft/f-13-docs-trans-schedu...

This version is a combination of:

1. key milestones for the release as a whole (to provide context)
2. Documentation tasks
3. Translation tasks

The most helpful feedback you can give me (at the meeting or elsewhere) is to provide corrections, additions, or deletions based on the version I've created. It does not help me if you create your own version from scratch and refer to it because then I have to start all over which is a big investment of time. A patch or diff against the original file works too.

Thanks, John

-- Fedora-trans-list mailing list Fedora-trans-list@redhat.com https://www.redhat.com/mailman/listinfo/fedora-trans-list

On 12/15/2009 04:24 PM, John J. McDonough wrote:

On Tue, 2009-12-15 at 15:05 -0800, John Poelstra wrote:

No question on that, but looking at that whenisgood is pretty discouraging. Monday came up too fast but next week is Christmas which kills a big chunk of the week. What is the matter with later this week or early next?

oops... that's what I meant to do.

I'll have to resend.

John

On 12/17/2009 05:32 PM, John Poelstra wrote:

On 12/15/2009 04:24 PM, John J. McDonough wrote:

...

On Tue, 2009-12-15 at 15:05 -0800, John Poelstra wrote:

...

I'll try to see what I can do with this, but it doesn't address my original request below.

Yeah, but 90% of it doesn't matter.

It really doesn't make a lot of difference whether you arrange the furniture or hang the pictures first, but you don't want to hang the pictures before you paint the wall, and oh yeah, maybe you need to put up the wall before you can paint it.

In F12 there was so much clutter that we missed some key items. That's why I'm focusing on getting the house built before I worry about decorating it.

Okay. This is the first time I've heard this so as you can imagine it's kind of hitting me out of the blue. Are you wanting to start the schedule from scratch and thus remove the parts you consider "clutter?"

Can you clarify what you expect our starting point to be at Tuesday's meeting so that I'm coming with the right expectations? (choose one) a) Blank slate b) Re-work draft I posted c) Your version

...

John, are these all part of a master schedule with dependencies between the teams, or is each team's plan independent? If we can just get the dependencies right we ought to have the dates right.

Both. Most of the docs tasks depend on each other, but often when they start is depended on other teams or schedule milestones. We've been trying to get the dependencies correct for a few releases... each time we're (hopefully) getting a little bit closer.

Thanks, John

I'm still not clear what our starting point for tomorrow night's meeting is. Since I haven't seen a response (forgive me if I missed one) I'm assuming we will be using (#b):

http://poelstra.fedorapeople.org/schedules/f-13-draft/f-13-docs-trans-schedu...

I have already put it on gobby. Feel free to start editing gobby now.

Note, the names of some of the releng milestones (Beta Freeze and Final Freeze) changed to match: https://fedoraproject.org/wiki/User:Poelstra/Key_Milestones

What can really help make tomorrow's meeting tremendously faster is to populate the drafted version on gobby in advance with what you think the dates should be to address the concerns you raised at: https://www.redhat.com/archives/fedora-docs-list/2009-December/msg00040.html Then we we can spend most of our time critiquing the draft instead of going line by line which will take a lot more time.

If you'd like to call out task dependencies, maybe add that as part of the task name in ( ).

I spent some more time today trying to understand the issues you are raising in the email and http://jjmcd.fedorapeople.org/F13-RelNotes-MajorTasks.html but it is not completely clear to me if these are the real dates you want to use or if this was just a "brainstorm."

Thanks, John

Sorry, pls ignore. I got the answer by myself... http://fedoraproject.org/wiki/Communicate/GobbyHowTo#Get_the_Password

Noriko Mizumoto さんは書きました:

<snip> > I have already put it on gobby. Feel free to start editing gobby now.

The password I used for F12 is not working anymore... Could you please guide me how to obtain one?

noriko

...

Note, the names of some of the releng milestones (Beta Freeze and Final Freeze) changed to match: https://fedoraproject.org/wiki/User:Poelstra/Key_Milestones

What can really help make tomorrow's meeting tremendously faster is to populate the drafted version on gobby in advance with what you think the dates should be to address the concerns you raised at: https://www.redhat.com/archives/fedora-docs-list/2009-December/msg00040.html Then we we can spend most of our time critiquing the draft instead of going line by line which will take a lot more time.

If you'd like to call out task dependencies, maybe add that as part of the task name in ( ).

I spent some more time today trying to understand the issues you are raising in the email and http://jjmcd.fedorapeople.org/F13-RelNotes-MajorTasks.html but it is not completely clear to me if these are the real dates you want to use or if this was just a "brainstorm."

Thanks, John

On Mon, 2009-12-21 at 10:57 -0800, John Poelstra wrote:

I'm still not clear what our starting point for tomorrow night's meeting is. Since I haven't seen a response (forgive me if I missed one) I'm assuming we will be using (#b):

http://poelstra.fedorapeople.org/schedules/f-13-draft/f-13-docs-trans-schedu...

I have already put it on gobby. Feel free to start editing gobby now.

I had kind of hoped that you would fix the dependencies on your schedule. We are still building rpms later than we need them.

Note, the names of some of the releng milestones (Beta Freeze and Final Freeze) changed to match: https://fedoraproject.org/wiki/User:Poelstra/Key_Milestones

What can really help make tomorrow's meeting tremendously faster is to populate the drafted version on gobby in advance with what you think the dates should be to address the concerns you raised at: https://www.redhat.com/archives/fedora-docs-list/2009-December/msg00040.html Then we we can spend most of our time critiquing the draft instead of going line by line which will take a lot more time.

I'm not nearly as concerned with the dates as with the dependencies.

If you'd like to call out task dependencies, maybe add that as part of the task name in ( ).

I spent some more time today trying to understand the issues you are raising in the email and http://jjmcd.fedorapeople.org/F13-RelNotes-MajorTasks.html but it is not completely clear to me if these are the real dates you want to use or if this was just a "brainstorm."

Again, I'm not nearly as concerned with the dates as with the dependencies. I've put some comments in the gobby doc, but it is pretty hard to follow in that format.

My main concerns are: - Build RPM before RelEng needs it, not after - Translate before build RPM - Write before translate

We seem to have the second two down, but the gobby sked still has us building RPMs after RelEng needs them.

I do think we need a few days for the beta RPM, and probably the RC RPM. It might make sense to make an RPM for RelEng's test compose, and leave say 3 days for that, then make another RPM for the final compose. That one we could do in a day. It only takes a few hours to make an RPM, but we always seem to run into some sort of issue. I figure if we deal with those issues for test compose, we can allow translation to continue and when we build for final compose we will have dealt with any issues that arise.

We also don't quite understand what our interface with L10N looks like this go. there are a number of issues we need to work there:

- Is the new Transifex finally there? - If not, does L10N do the POT merging/PO splitting? - Do we do a ton of branches like we did for F12? - Does L10N make their own htmls as needed? - Does L10N do the push to docs.fp.o?

--McD

On Tue, 2009-12-22 at 10:15 -0800, John Poelstra wrote:

What do you mean by "fixing the dependencies?"

"Build fed-rel-notes.rpm for Beta Release" (25) needs to have as a successor, "Compose Beta Candidate"

Similarly, "Build fedora-release-notes.rpm" (50) needs to be a predecessor for "Create RC Test Compose (TC)"

I understood from the beginning that the dates were wrong, but I was looking to you to supply the correct dates.

Absolute dates are meaningless. What is important is that the RPM get built BEFORE releng does the compose

It would help me a lot if you could be extremely specific with what you are needing or expecting from me. Something like, "John, please take the dates for the release notes from the bottom of the html page I created and replace tasks #__ to #__ on the schedule you created with them. Have the releases notes start/depend on task #___ from release engineering."

Make the RC test compose a successor to task 50, and the beta compose a successor to task 50

I'd like to understand this better. What do you mean by "the dependencies?" What problem are you trying to solve by seeing these dependencies?

The problem is that the Fedora 12 schedule had Docs building the RPMs AFTER releng needed them. I just want us to have an RPM in hand at the right time.

To me the dependencies are inherent in the dates you propose.... one task follows another and can't start unless the previous task was completed. As set forth each task with its start and end date, I'm assuming this is being taken into account.

The dates ARE A RESULT OF the dependencies.

All this focus on the "dependencies" is new to me. We've done several previous detailed reviews of the docs schedule and you are the first to raise this as such an important issue. I want to help, but I'm failing to see how to provide you this information or what problem you're trying to solve.

It can't be that new to you -- there are a bazillion of them in your .tjp file.

I haven't found an easy way to review and create schedules virtually. In person we could have probably finished all of this in an hour. It is tedious and really not that fun of task. Sometimes I have to print a schedule off, turn off my computer screen to fully concentrate and walk it through with a calendar nearby.

Hey, there's no easy way to review and create schedules virtually or otherwise.

What format would be more helpful for you?

I was hoping that I could take your TaskJuggler file and maybe put it in that language, but alas, it is just too opaque for me.

Yes, I'm waiting for you to insert whatever you want the dates to be.

Please insert/change whatever you think the dates should be.

Dates, dates, dates. I don't care what the date is, only that it is the right date relative to releng.

Thanks for being patient with me. I really do want to find a better way to do this :)

Well, we have sixty three things going on at once, and I know that sometimes I'm not the easiest person in the world to understand, so I certainly appreciate your patience.

I hate to massively hack up your schedule, although I don't think it actually is a huge hack; just a couple of dependencies and assuming the other dependencies are in place (and they seem to be although I didn't go through the tjp line by line), it should be simply adding a couple of dependencies and let the software do its thing.

--McD

----- Original Message ----- From: "Paul W. Frields" stickster@gmail.com Sent: Friday, December 18, 2009 3:55 PM Subject: Re: Documentation & Translation schedule review

to do so. It's well-nigh impossible to relegate schedule data from completely different tools.

I agree. I was just using the quickest way to check that key items on John's list made sense. That's why I didn't take time to schedule all the peripheral tasks. Yes, we need to make sure they get done, but the web output is very hard to really understand. (Not that the web output from planner is all that much better!). It is very hard to see whether the key items are being appropriately handled. A schedule with just the key items helps see that. Even with just a handful of tasks, the Gantt chart is just too long and gangly to digest.

In Fedora 12, the schedule had us making the release notes rpm TWO WEEKS after RelEng needed it. I wanted to avoid that this time. Heck of it is, I spent a lot of time with John on that schedule, and paid pretty close attention to it, and never caught that little detail until the last minute. Fortunately I did catch it, barely in time.

I suspect the talk next week will help a lot in this regard. I'd also

And that really is the point. I need that visibility into the dependencies to make sure we don't miss the key items in our zeal to make sure we have all the details covered.

recommend that people on the team think about how we could use the resulting schedule in regular meetings to stay on top of approaching tasks.

The problem was that we WERE paying attention to the schedule, and missing the larger problem. I want to make sure we get the schedule right this time. When we get in the midst of it, we really can't do anything but watch the schedule. What's the saying, when you are ass deep in alligators it is hard to keep in mind your objective is to drain the swamp? We need that roadmap, but we don't want it to get us to the station after the train has left.

There's always a lot to do, and I really appreciate John doing the heavy lifting here and making sure all the details are taken care of. If I were doing it, it would be an unmitigated disaster. But I do want to emphasize that there are a few very key tasks that drive the rest of the schedule, and most importantly, they are ultimately tied to RelEng. We have the most touch points with L10N, but the key, date-determining connections are with RelEng, and somehow those got missed last time.

--McD