evolve: don't use graftstate for continuing `hg evolve` Before this patch, if user runs `hg evolve`, we read the evolve state, write a graftstate from that information and run graft command. This patch replaces that logic with new logic which does not depends on the graft state. The new logic is very filtered part of logic from graft command. We did not require a lot of opts checking and filtering revs logic from the graft command as our use case is fixed. This is the start of series/work which will make `hg evolve` better and have a good statefile for itself.

topics: add a new templatekeyword `topic` This patch adds a new templatekeyword topic which can be used to list the topic of the changeset when using commands which supports templating.

packaging: update MANIFEST.in to include new tutorial files I forget to update the MANIFEST.in file when important the new tutorial files. This commit fix that.

test-compat: merge mercurial-4.2 into mercurial-4.1

test-compat: merge mercurial-4.3 into mercurial-4.2

test-compat: merge stable into mercurial-4.3

test: add a test to reproduce issue 5720

test: add missing cycle test Also fix precursors to break on obs markers cycles.

Small typos fixing in the topic tutorial

doc: import the training support Import the training support which was stored in a private-repository before.

branching: merge with stable

evolve: use new trouble names in commit messages There is case when we add trouble to the commit message while using `hg evolve`. This match makes sure, we use the new trouble names. The hash changed because it depends on commit messages.

evolve: use compat.TROUBLES to show troubles in user interface compat.TROUBLES is a dict which contains the trouble name for each trouble according to the hg version user is using. All of the replacements are related to `hg evolve` command.

topics: list only topic names in quiet mode This will be helpful in bash completion.

topics: take logic to parse username to a separate function In next patch we will be adding support to `hg stack` to show users, and this logic will be required there too. So let's take it out in a separate function.

changelog: fix a typo in the changelog There is probably many others but that one is fixed now.

caches: switch to 'auto' warming by default This options seems good let us make it the default.

caches: add a 'auto' option for obshashrange cache warming This option will only warm the cache when used as a server.

caches: extract some config reading in 'shouldwarmcache'

caches: pass the transaction to the "shouldwarncache" logic This will allow to have smarter mode about cache warming (eg: only warm them for server transaction.

caches: record 'desc' attribute on transaction This is useful to know if a transaction if server side or not.

caches: protect against missing connection If the connection cannot be established, we should not try to use it.

evolve: don't show working directory obsolete message if we were on it This patch tweaks showing wc obsolete message functionality to not show the message if the operation led us to the same changeset we were on before and it didn't obsoleted it. This make failed updates, update on current changeset and a pull with no change omit showing the message. This has some cons like if you are on rev 1 which is obsolete and you do `hg up <revset>` where revset is some revset which resolves to the rev 1, we won't show the warning as shown earlier.

prev: jump to parent's successor if parent is obsolete and topic is involved `hg stack` shows a linear chain of commits even when they are not linear chain and some changesets are obsoleted and have successors as separate head. This is very nice to have a post evolution view of the stack. However when `hg prev` is run on an unstable changeset whose parent is obsolete and does not topic but parents successor has the topic, it shows no parent on the topic. This patch makes `hg prev` update to parent's successor and make things follow the stack order.

tests: add a test showing wrong behaviour doing `hg prev` on orphan cset If there is an unstable changeset in the topic stack, whose parent is not a part of stack but has a successor which is part of stack, hg stack does shows them in linear order, but `hg prev` on that orphan changset says no parent on that topic. However it should update to the successor of the changeset. It will be fixed in the next patch.

test: update output to 02fdb8c018aa Now that we have directaccess in core, we get a new warning message when updating to a hidden changeset. Update the tests. CORE-TEST-OUTPUT-UPDATE: 02fdb8c018aa

test: update output to 9b3f95d9783d The unstable graph nodes are now shown with the character "*", update the test outputs. CORE-TEST-OUTPUT-UPDATE: 9b3f95d9783d

stablesort: use parent filtering in a place we forgot to Freak merge striked again.

obshashrange: add a progressbar to upgrade This section is the slowest, have some progress idea would be useful.

stablerange: add an sql index on subranges id too This should speed up query that find super ranges.

obshashrange: less brutal reset when receiving markers on existing node We now target the affected ranges only. The implementation is still slow but that should be better than what we had before.

obshashrange: update the cache updating logic Use the more modern option when available. (Same as the other caches).

stablerange: cleanup the update logic

stablesort: cleanup the update logic

firstmergecache: cleanup the update logic

depthcache: cleanup the update logic

changelog: add an entry about the obshashrange changes

stablerange: be more cautious when deleting the sql connection This triggered some crash in further development. This also match what is done elsewhere with this attribute.

stablerange: clarify sql transaction logic This should help concurrent access.

stablerange: use mergepoint based algorithm for the official stable range Changing the official stable range will impact all users of the infrastructure. We update version number of the cache file and discovery methods to clarify this. We do no keep compatibility with the old method over the wire. The new algorithm is much better and keeping compat is more work than we have time for right now.

stablerange: explicitly create a stablerange entry for branchpoint method We are about to change the default.

stablerange: have a mergepoint based sql backed class Everything is so generic that getting a class is 3 line of codes.

stablerange: introduce a sql backed version of the generic on disk cache This is getting use close to actually using the new sorting and algorithm for real.

stablerange: add a base class for on disk stored stablerange class

stablesort: use on disk cache for test

caches: factorise the cache warming check The old conditional was duplicated and hard to read. We factorise everything into the `utility` module.

stablerange: use repo-carried stablesortcache That one is common to all and disk persisted

stablesort: expose the cache through the repository Let have a unique instance, keep it warm and accessible.

stablesort: implement an ondisk cache Persisting the cache on disk will be important for large repositories.

stablesort: realign a misaligned continue We should make a better use of the jump cache now.

stablesort: remove some dead code This closure is no longer in use.

stablesort: abstract all cache access This prepare an on disk version of the cache

stablerange: use first merge cache to skip over linear section If we are to skip over a large section of the range, skipping everything until the first merge seems like a good idea.

stablerange: add a new 'firstmerge' cache This cache store the first merge ancestors of a changesets. This is useful to avoid iteration over the changelog when building stablerange.

stablerange: drop unused `until` utility We no longer needs this function.

stablerange: use cached size data instead of walking the graph Less iteration is better.

stablesort: record previous segment size in the jump That was, we can skip over a full jump without walking it. We'll make use of this in the next changeset.

stablesort: move jump recording inside the exclusive function This will allow use to record to cache size of segment in a later changeset. If the exclusive set is empty, we need to record it outside that function.

stablerange: compute jump size after jump retrieval only This prepares the caching of jump sizes.

stablesort: warm jump cache more efficiently We no longer for a full depth iteration for all request: - we exit early for non-merge, - for merge, we only iterate over the exclusive area.

stablesort: use a regular dict for jumps This will help knowing what revision we have already queried or not.

stablerange: use the jump information for faster iteration Instead of doing a full iteration over the exclusive set, we compare the jump information instead. This perform the same kind of logic than before but will allow for much faster iteration once all the caches are in place.

stablesort: expose the jumps sequence to other code The stable range needs it to build exclusive subrange efficiently.

stablesort: use 'depth' in mergepoint tie breaker The parents with the most depth will is considered lower. It has a couple of advantages. 1) the more shallow parent probably have less exclusive revision, 2) it makes Oedipus merge behave like close to the linear case,