The `draft() & ::x` type query could be common for selecting one or more
draft feature branches being worked on.
Before this patch, `::x` may travel through the changelog DAG for a long
distance until it gets a smaller revision number than `min(draft())`. It
could be very slow on long changelog with distant (in terms of revision
numbers) drafts.
This patch adds a fast path for this situation, and will stop traveling the
changelog DAG once `::x` hits a non-draft revision.
The fast path also works for `secret()` and `not public()`.
To measure the performance difference, I used drawdag to create a repo that
emulates distant drafts:
DRAFT4
|
DRAFT3 # draft
/
PUBLIC9999 # public
|
PUBLIC9998
|
. DRAFT2
. |
. DRAFT1 # draft
| /
PUBLIC0001 # public
And measured the performance using the repo:
(BEFORE)
$ hg perfrevset 'draft() & ::(DRAFT2+DRAFT4)'
! wall 0.017132 comb 0.010000 user 0.010000 sys 0.000000 (best of 156)
$ hg perfrevset 'draft() & ::(all())'
! wall 0.024221 comb 0.030000 user 0.030000 sys 0.000000 (best of 113)
(AFTER)
$ hg perfrevset 'draft() & ::(DRAFT2+DRAFT4)'
! wall 0.000243 comb 0.000000 user 0.000000 sys 0.000000 (best of 9303)
$ hg perfrevset 'draft() & ::(all())'
! wall 0.004319 comb 0.000000 user 0.000000 sys 0.000000 (best of 655)
Differential Revision: https://phab.mercurial-scm.org/D441
This should make optimize() more readable and less error-prone, but it doubles
the parsing cost.
(original)
$ python -m timeit -n10000 -s 'from mercurial import revsetlang as L' \
'L.optimize(L.analyze(L.parse("ancestors(x) and not ancestors(y)")))'
10000 loops, best of 3: 79.3 usec per loop
(this patch)
$ python -m timeit -n10000 -s 'from mercurial import revsetlang as L' \
'L._treecache.clear(); \
L.optimize(L.analyze(L.parse("ancestors(x) and not ancestors(y)")))'
10000 loops, best of 3: 201 usec per loop
This should make optimize() more readable, but it doubles the parsing cost.
(original)
$ python -m timeit -n10000 -s 'from mercurial import revsetlang as L' \
'L.optimize(L.analyze(L.parse("::tip")))'
10000 loops, best of 3: 18.1 usec per loop
(this patch)
$ python -m timeit -n10000 -s 'from mercurial import revsetlang as L' \
'L._treecache.clear(); L.optimize(L.analyze(L.parse("::tip")))'
10000 loops, best of 3: 48.4 usec per loop
30usec isn't dominant compared to the revset evaluation, but that is a cost.
That's why a parsed tree is cached, which can benefit in hgweb or chg server.
Rewrite `flipand(y, x)` to `andsmally(x, y)` so the AST order is unchanged,
which could be more friendly to developers.
Differential Revision: https://phab.mercurial-scm.org/D579
The old documentation is a bit confusing. Namely, it's unclear whether
`define` means "I should ALWAYS define a new order", or "I should SOMETIMES
define a new order", and if it's the latter, what's the difference between
`define` and `any`?
This patch clarifies that and adds more examples.
Differential Revision: https://phab.mercurial-scm.org/D523
Keeping `order` in tree makes AST operation harder. And there could be
invalid cases if trees could be generated and compounded freely, like:
SetA(order=define) & SetB(order=define)
^^^^^^ couldn't be satisfied
This patch changes the code to calculate order on the fly, during tree
traversal. Optimization of reordering `and` arguments is preserved by
introducing a new internal operation `flipand`.
.. api::
revset.stringset() now takes 'order' as the last argument.
Differential Revision: https://phab.mercurial-scm.org/D451
The reordering optimization is more important for "and" than "or", given the
implementation details about "addset" and "filteredset" - reordering "or"
may help "__contains__" test but not iteration, reordering "and" could help
both. We are going to simplify the tree to remove ordering information.
Removing "or" reordering optimization would make things simpler.
This effectively reverts 6820a8a645ef. It tracks back to the "orset"
function added by the initial commit of revset (c9ce8ecd6).
In the future, we might consider optimization at runtime (ex. do reordering
and rewrites inside "orset").
Differential Revision: https://phab.mercurial-scm.org/D561
The proposed syntax [1] was originally 'set{n rel}', but it seemed slightly
confusing if template is involved. On the other hand, we want to keep 'set[n]'
for future extension. So this patch introduces 'set#rel[n]' ternary operator.
I chose '#' just because it looks like applying an attribute.
This also adds stubs for 'set[n]' and 'set#rel' operators since these syntax
elements are fundamental for constructing 'set#rel[n]'.
[1]: https://www.mercurial-scm.org/wiki/RevsetOperatorPlan#ideas_from_mpm
Previously repo.anyrevs only expand aliases in [revsetalias] config. This
patch makes it more flexible to accept a customized dict defining aliases
without having to couple with ui.
revsetlang.expandaliases now has the signature (tree, aliases, warn=None)
which is more consistent with templater.expandaliases. revsetlang.py is now
free from "ui", which seems to be a good thing.
New revsetlang module hosts parser, tokenizer, and miscellaneous functions
working on parsed tree. It does not include functions for evaluation such as
getset() and match().
2288 mercurial/revset.py
684 mercurial/revsetlang.py
2972 total
get*() functions are aliased since they are common in revset.py.