1.7 Developer Sprint

This was a long weekend of in-depth Mercurial discussion, design, and coding. The sprint took place October 8th-10th 2010 at Google Chicago.

Sprints are open to all significant Mercurial contributors. See 1.5sprint for history.

1. Attendees

2. What happened at the sprint?

Here's a mostly complete overview of features, discussions, bugs, patches, etc. that came up (from a posting to mercurial-devel by Brodie):

2.1. Big new features

• Online help in hg serve/hgweb (durin42)

• Parentdelta support (reduces repo size) (not yet enabled by default) (tonfa)

2.2. Unfinished business

• Greatly improved changeset discovery (speeds up push/pull) (tonfa)

• "LiquidHg" - a system for safely allowing mutable history (mpm)

• "Dead heads" support for marking changesets as abandoned (via pushkey) (mg, original email, patch queue)

• Better templater/templating language

• Revset support everywhere -r/--rev is used (mpm)

• "Black box" extension that records warnings, errors, etc. (nicdumz)
• Improved extension error messages (nicdumz)

• Test potential migration from Roundup to http://trac.edgewall.org/ for bug tracking (bmp, brodie)

• No prompts from "hg merge" (use hg resolve to handle merges at your own pace instead) (muggs)

• Automated PyPI upload through setup.py (tonfa, brodie)

• hg config --edit (brodie, kiilerix, mg)
• More consistent/reliable internal URL parsing (brodie, kiilerix)

• Fine-grained HGPLAIN (e.g., make stuff plain except i18n) (brodie)

• subrepo and non-default pull/push paths, issue1852, patch (kiilerix, mg)

2.3. Discussions

• Cleaning up and reorganizing the wiki
• Scaling hgweb (smarter etags, better caching)

• Improving Google Summer of Code (mpm, brodie, durin42, tonfa, nicdumz, mg)

• Improving extension debugging (extensions could provide homepage URLs, their own version numbers, and specify what versions of Mercurial they support)

• Better changeset signing support (support for more tools, commitsigs integration, pushkey-based changeset signing)

• Moving extdiff into core and making it use merge-tools settings

• Moving other extensions into core (progress, record, transplant, graphlog)

• Better auto-generated documentation (via Sphinx, epydoc, etc.)

• Adding more command examples in help

• Better documenting how --verbose/--quiet affects different commands

• Better documenting internal:* merge settings

• Specifying merge tools on the command line

• Improving the release process

• Improving subrepo support (should we use it ourselves? polib is a candidate)

2.4. Patches for bugs

• issue2364 - merge: handle no file parent in backwards merge (mpm)

• issue2373 - ui.paths: expand paths directly in fixconfig (tonfa)

• issue2427 - mq: handle deleting the same patch twice in one command (danchr)

• issue2402 - mq: handle qfolding empty patches (pmezard)

• issue1388 - patch: disable patcher fallback if ui.patch is not set (pmezard)

• issue2191 - progress: make sure stderr has isatty before calling (durin42)

• issue2074 - windows: handle spaces in path to Python (bmp)

• issue2417 - import: don't strip '#' lines from patch descriptions (kiilerix)

• issue2376 - dispatch: properly handle relative path aliases used with -R (brodie)

• issue2361 - revset: lower precedence of minus infix (mpm)

• issue1564 - strip: add --keep flag to avoid modifying wc during strip (durin42)

2.5. Other interesting patches that made it in

• setup/hg: always load Mercurial from where it was installed. (10da5a1f25dd) (danchr)
• setup: user-friendly error message if Python headers are missing (6c0e1aee1b19) (nicdumz)

• pager: don't run pager if nothing is written to stdout/stderr (05077896ffe2) (brodie)

• pager: add global --pager=<auto/boolean> option (04f6de46bf3a) (brodie)

• copies: don't detect copies as "divergent renames" (ada47c38f4e5) (danchr)

2.6. Miscellaneous

• Made mercurial-consulting mailing list (mpm)

2.7. See also

• http://blog.bitbucket.org/2010/10/09/mercurial-1-7-sprint-in-chicago-ongoing/

• http://www.selenic.com/blog/?p=681

3. Information for attendees

As the sprint is over, this information is left for posterity.

3.1. Costs

We will fund hotel for all attendees. Where possible, attendees should try to get their employers to cover travel. We refund personal travel costs based on available funds and contribution level.

3.2. Date and Time

The sprint will be in Chicago on October 8th-10th 2010. This will allow us time to put some work started at the sprint into 1.7 before the code freeze on October 15th. Augie will be at Google Chicago starting around 8:30 on Friday, so feel free to start arriving anytime after that. Please try to arrive at least by 9:15 so we can get started promptly.

The Fairfield Inn hotel has been booked for three nights from October 7th to 10th -- checkin is Thursday, checkout is Sunday. You need to make other reservations if you arrive earlier or leave later.

Check-in on Thursday is at 3pm. The rooms are theoretically booked in your names, but if for some reason they can't find you, the reservation number is 15443446301.

3.3. Travel

Make sure you have your passports and visas figured out in advance. Plan to arrive the day before the sprint begins to be ready for a full day of workshop.

3.3.1. Getting to the Hotel from an Airport

From Midway: Take the Orange Line to Roosevelt. Transfer to the Red Line headed towards Howard. Get off the Red at Grand, walk two blocks North then turn right. You'll see the hotel on the left side of the street.

From O'Hare: Take the Blue Line to Monroe. Transfer to the Red Line headed towards Howard. Get off the Red at Grand, walk two blocks North then turn right. You'll see the hotel on the left side of the street. This is the Blue Line terminal in O'Hare: Farecard-Ohare.jpg.

This Hotel.jpg is what the hotel looks like from across the street.

For either public transit option from the airports, you'll need a farecard for the train. It's US$2.25 for either trip and transfers are free. The farecard machines in the airports do take cards as well as cash. Unless you plan to take the CTA for sightseeing while in town, there's no point in getting a 3-day pass, as you'll only need $4.50 of fares and the 3-day pass is $14.

If you plan to take a taxi, be aware that not all taxis accept credit cards. Many take only cash.

3.3.2. Getting to Google from the Hotel

Walk West on Ontario to State, turn left on State and go to Kinzie. Turn right on Kinzie. Google is in that block on the right, our entrance is just after the entrance to the Amalfi hotel. The entrance isn't marked as Google, but if you get to Dearborn Street or see Keifer's Steakhouse you're too far. There are two revolving doors, and a security guard. On Friday, there should be someone at a table (or something along those lines) to greet you just inside the door. Google.jpg is what Google looks like from across the street.

3.3.3. Hotel Arrangements

---

CategoryMeetings