##// END OF EJS Templates
wireproto: add streams to frame-based protocol...
wireproto: add streams to frame-based protocol Previously, the frame-based protocol was just a series of frames, with each frame associated with a request ID. In order to scale the protocol, we'll want to enable the use of compression. While it is possible to enable compression at the socket/pipe level, this has its disadvantages. The big one is it undermines the point of frames being standalone, atomic units that can be read and written: if you add compression above the framing protocol, you are back to having a stream-based protocol as opposed to something frame-based. So in order to preserve frames, compression needs to occur at the frame payload level. Compressing each frame's payload individually will limit compression ratios because the window size of the compressor will be limited by the max frame size, which is 32-64kb as currently defined. It will also add CPU overhead, as it is more efficient for compressors to operate on fewer, larger blocks of data than more, smaller blocks. So compressing each frame independently is out. This means we need to compress each frame's payload as if it is part of a larger stream. The simplest approach is to have 1 stream per connection. This could certainly work. However, it has disadvantages (documented below). We could also have 1 stream per RPC/command invocation. (This is the model HTTP/2 goes with.) This also has disadvantages. The main disadvantage to one global stream is that it has the very real potential to create CPU bottlenecks doing compression. Networks are only getting faster and the performance of single CPU cores has been relatively flat. Newer compression formats like zstandard offer better CPU cycle efficiency than predecessors like zlib. But it still all too common to saturate your CPU with compression overhead long before you saturate the network pipe. The main disadvantage with streams per request is that you can't reap the benefits of the compression context for multiple requests. For example, if you send 1000 RPC requests (or HTTP/2 requests for that matter), the response to each would have its own compression context. The overall size of the raw responses would be larger because compression contexts wouldn't be able to reference data from another request or response. The approach for streams as implemented in this commit is to support N streams per connection and for streams to potentially span requests and responses. As explained by the added internals docs, this facilitates servers and clients delegating independent streams and compression to independent threads / CPU cores. This helps alleviate the CPU bottleneck of compression. This design also allows compression contexts to be reused across requests/responses. This can result in improved compression ratios and less overhead for compressors and decompressors having to build new contexts. Another feature that was defined was the ability for individual frames within a stream to declare whether that individual frame's payload uses the content encoding (read: compression) defined by the stream. The idea here is that some servers may serve data from a combination of caches and dynamic resolution. Data coming from caches may be pre-compressed. We want to facilitate servers being able to essentially stream bytes from caches to the wire with minimal overhead. Being able to mix and match with frames are compressed within a stream enables these types of advanced server functionality. This commit defines the new streams mechanism. Basic code for supporting streams in frames has been added. But that code is seriously lacking and doesn't fully conform to the defined protocol. For example, we don't close any streams. And support for content encoding within streams is not yet implemented. The change was rather invasive and I didn't think it would be reasonable to implement the entire feature in a single commit. For the record, I would have loved to reuse an existing multiplexing protocol to build the new wire protocol on top of. However, I couldn't find a protocol that offers the performance and scaling characteristics that I desired. Namely, it should support multiple compression contexts to facilitate scaling out to multiple CPU cores and compression contexts should be able to live longer than single RPC requests. HTTP/2 *almost* fits the bill. But the semantics of HTTP message exchange state that streams can only live for a single request-response. We /could/ tunnel on top of HTTP/2 streams and frames with HEADER and DATA frames. But there's no guarantee that HTTP/2 libraries and proxies would allow us to use HTTP/2 streams and frames without the HTTP message exchange semantics defined in RFC 7540 Section 8. Other RPC protocols like gRPC tunnel are built on top of HTTP/2 and thus preserve its semantics of stream per RPC invocation. Even QUIC does this. We could attempt to invent a higher-level stream that spans HTTP/2 streams. But this would be violating HTTP/2 because there is no guarantee that HTTP/2 streams are routed to the same server. The best we can do - which is what this protocol does - is shoehorn all request and response data into a single HTTP message and create streams within. At that point, we've defined a Content-Type in HTTP parlance. It just so happens our media type can also work as a standalone, stream-based protocol, without leaning on HTTP or similar protocol. Differential Revision: https://phab.mercurial-scm.org/D2907

File last commit:

r37022:68fcc550 default
r37304:9bfcbe4f default
Show More
destutil.py
415 lines | 15.2 KiB | text/x-python | PythonLexer
# destutil.py - Mercurial utility function for command destination
#
# Copyright Matt Mackall <mpm@selenic.com> and other
#
# This software may be used and distributed according to the terms of the
# GNU General Public License version 2 or any later version.
from __future__ import absolute_import
from .i18n import _
from . import (
bookmarks,
error,
obsutil,
scmutil,
stack
)
def _destupdateobs(repo, clean):
"""decide of an update destination from obsolescence markers"""
node = None
wc = repo[None]
p1 = wc.p1()
movemark = None
if p1.obsolete() and not p1.children():
# allow updating to successors
successors = obsutil.successorssets(repo, p1.node())
# behavior of certain cases is as follows,
#
# divergent changesets: update to highest rev, similar to what
# is currently done when there are more than one head
# (i.e. 'tip')
#
# replaced changesets: same as divergent except we know there
# is no conflict
#
# pruned changeset: no update is done; though, we could
# consider updating to the first non-obsolete parent,
# similar to what is current done for 'hg prune'
if successors:
# flatten the list here handles both divergent (len > 1)
# and the usual case (len = 1)
successors = [n for sub in successors for n in sub]
# get the max revision for the given successors set,
# i.e. the 'tip' of a set
node = repo.revs('max(%ln)', successors).first()
if bookmarks.isactivewdirparent(repo):
movemark = repo['.'].node()
return node, movemark, None
def _destupdatebook(repo, clean):
"""decide on an update destination from active bookmark"""
# we also move the active bookmark, if any
activemark = None
node, movemark = bookmarks.calculateupdate(repo.ui, repo, None)
if node is not None:
activemark = node
return node, movemark, activemark
def _destupdatebranch(repo, clean):
"""decide on an update destination from current branch
This ignores closed branch heads.
"""
wc = repo[None]
movemark = node = None
currentbranch = wc.branch()
if clean:
currentbranch = repo['.'].branch()
if currentbranch in repo.branchmap():
heads = repo.branchheads(currentbranch)
if heads:
node = repo.revs('max(.::(%ln))', heads).first()
if bookmarks.isactivewdirparent(repo):
movemark = repo['.'].node()
elif currentbranch == 'default' and not wc.p1():
# "null" parent belongs to "default" branch, but it doesn't exist, so
# update to the tipmost non-closed branch head
node = repo.revs('max(head() and not closed())').first()
else:
node = repo['.'].node()
return node, movemark, None
def _destupdatebranchfallback(repo, clean):
"""decide on an update destination from closed heads in current branch"""
wc = repo[None]
currentbranch = wc.branch()
movemark = None
if currentbranch in repo.branchmap():
# here, all descendant branch heads are closed
heads = repo.branchheads(currentbranch, closed=True)
assert heads, "any branch has at least one head"
node = repo.revs('max(.::(%ln))', heads).first()
assert node is not None, ("any revision has at least "
"one descendant branch head")
if bookmarks.isactivewdirparent(repo):
movemark = repo['.'].node()
else:
# here, no "default" branch, and all branches are closed
node = repo.lookup('tip')
assert node is not None, "'tip' exists even in empty repository"
return node, movemark, None
# order in which each step should be evaluated
# steps are run until one finds a destination
destupdatesteps = ['evolution', 'bookmark', 'branch', 'branchfallback']
# mapping to ease extension overriding steps.
destupdatestepmap = {'evolution': _destupdateobs,
'bookmark': _destupdatebook,
'branch': _destupdatebranch,
'branchfallback': _destupdatebranchfallback,
}
def destupdate(repo, clean=False):
"""destination for bare update operation
return (rev, movemark, activemark)
- rev: the revision to update to,
- movemark: node to move the active bookmark from
(cf bookmark.calculate update),
- activemark: a bookmark to activate at the end of the update.
"""
node = movemark = activemark = None
for step in destupdatesteps:
node, movemark, activemark = destupdatestepmap[step](repo, clean)
if node is not None:
break
rev = repo[node].rev()
return rev, movemark, activemark
msgdestmerge = {
# too many matching divergent bookmark
'toomanybookmarks':
{'merge':
(_("multiple matching bookmarks to merge -"
" please merge with an explicit rev or bookmark"),
_("run 'hg heads' to see all heads")),
'rebase':
(_("multiple matching bookmarks to rebase -"
" please rebase to an explicit rev or bookmark"),
_("run 'hg heads' to see all heads")),
},
# no other matching divergent bookmark
'nootherbookmarks':
{'merge':
(_("no matching bookmark to merge - "
"please merge with an explicit rev or bookmark"),
_("run 'hg heads' to see all heads")),
'rebase':
(_("no matching bookmark to rebase - "
"please rebase to an explicit rev or bookmark"),
_("run 'hg heads' to see all heads")),
},
# branch have too many unbookmarked heads, no obvious destination
'toomanyheads':
{'merge':
(_("branch '%s' has %d heads - please merge with an explicit rev"),
_("run 'hg heads .' to see heads")),
'rebase':
(_("branch '%s' has %d heads - please rebase to an explicit rev"),
_("run 'hg heads .' to see heads")),
},
# branch have no other unbookmarked heads
'bookmarkedheads':
{'merge':
(_("heads are bookmarked - please merge with an explicit rev"),
_("run 'hg heads' to see all heads")),
'rebase':
(_("heads are bookmarked - please rebase to an explicit rev"),
_("run 'hg heads' to see all heads")),
},
# branch have just a single heads, but there is other branches
'nootherbranchheads':
{'merge':
(_("branch '%s' has one head - please merge with an explicit rev"),
_("run 'hg heads' to see all heads")),
'rebase':
(_("branch '%s' has one head - please rebase to an explicit rev"),
_("run 'hg heads' to see all heads")),
},
# repository have a single head
'nootherheads':
{'merge':
(_('nothing to merge'),
None),
'rebase':
(_('nothing to rebase'),
None),
},
# repository have a single head and we are not on it
'nootherheadsbehind':
{'merge':
(_('nothing to merge'),
_("use 'hg update' instead")),
'rebase':
(_('nothing to rebase'),
_("use 'hg update' instead")),
},
# We are not on a head
'notatheads':
{'merge':
(_('working directory not at a head revision'),
_("use 'hg update' or merge with an explicit revision")),
'rebase':
(_('working directory not at a head revision'),
_("use 'hg update' or rebase to an explicit revision"))
},
'emptysourceset':
{'merge':
(_('source set is empty'),
None),
'rebase':
(_('source set is empty'),
None),
},
'multiplebranchessourceset':
{'merge':
(_('source set is rooted in multiple branches'),
None),
'rebase':
(_('rebaseset is rooted in multiple named branches'),
_('specify an explicit destination with --dest')),
},
}
def _destmergebook(repo, action='merge', sourceset=None, destspace=None):
"""find merge destination in the active bookmark case"""
node = None
bmheads = bookmarks.headsforactive(repo)
curhead = repo[repo._activebookmark].node()
if len(bmheads) == 2:
if curhead == bmheads[0]:
node = bmheads[1]
else:
node = bmheads[0]
elif len(bmheads) > 2:
msg, hint = msgdestmerge['toomanybookmarks'][action]
raise error.ManyMergeDestAbort(msg, hint=hint)
elif len(bmheads) <= 1:
msg, hint = msgdestmerge['nootherbookmarks'][action]
raise error.NoMergeDestAbort(msg, hint=hint)
assert node is not None
return node
def _destmergebranch(repo, action='merge', sourceset=None, onheadcheck=True,
destspace=None):
"""find merge destination based on branch heads"""
node = None
if sourceset is None:
sourceset = [repo[repo.dirstate.p1()].rev()]
branch = repo.dirstate.branch()
elif not sourceset:
msg, hint = msgdestmerge['emptysourceset'][action]
raise error.NoMergeDestAbort(msg, hint=hint)
else:
branch = None
for ctx in repo.set('roots(%ld::%ld)', sourceset, sourceset):
if branch is not None and ctx.branch() != branch:
msg, hint = msgdestmerge['multiplebranchessourceset'][action]
raise error.ManyMergeDestAbort(msg, hint=hint)
branch = ctx.branch()
bheads = repo.branchheads(branch)
onhead = repo.revs('%ld and %ln', sourceset, bheads)
if onheadcheck and not onhead:
# Case A: working copy if not on a head. (merge only)
#
# This is probably a user mistake We bailout pointing at 'hg update'
if len(repo.heads()) <= 1:
msg, hint = msgdestmerge['nootherheadsbehind'][action]
else:
msg, hint = msgdestmerge['notatheads'][action]
raise error.Abort(msg, hint=hint)
# remove heads descendants of source from the set
bheads = list(repo.revs('%ln - (%ld::)', bheads, sourceset))
# filters out bookmarked heads
nbhs = list(repo.revs('%ld - bookmark()', bheads))
if destspace is not None:
# restrict search space
# used in the 'hg pull --rebase' case, see issue 5214.
nbhs = list(repo.revs('%ld and %ld', destspace, nbhs))
if len(nbhs) > 1:
# Case B: There is more than 1 other anonymous heads
#
# This means that there will be more than 1 candidate. This is
# ambiguous. We abort asking the user to pick as explicit destination
# instead.
msg, hint = msgdestmerge['toomanyheads'][action]
msg %= (branch, len(bheads) + 1)
raise error.ManyMergeDestAbort(msg, hint=hint)
elif not nbhs:
# Case B: There is no other anonymous heads
#
# This means that there is no natural candidate to merge with.
# We abort, with various messages for various cases.
if bheads:
msg, hint = msgdestmerge['bookmarkedheads'][action]
elif len(repo.heads()) > 1:
msg, hint = msgdestmerge['nootherbranchheads'][action]
msg %= branch
elif not onhead:
# if 'onheadcheck == False' (rebase case),
# this was not caught in Case A.
msg, hint = msgdestmerge['nootherheadsbehind'][action]
else:
msg, hint = msgdestmerge['nootherheads'][action]
raise error.NoMergeDestAbort(msg, hint=hint)
else:
node = nbhs[0]
assert node is not None
return node
def destmerge(repo, action='merge', sourceset=None, onheadcheck=True,
destspace=None):
"""return the default destination for a merge
(or raise exception about why it can't pick one)
:action: the action being performed, controls emitted error message
"""
# destspace is here to work around issues with `hg pull --rebase` see
# issue5214 for details
if repo._activebookmark:
node = _destmergebook(repo, action=action, sourceset=sourceset,
destspace=destspace)
else:
node = _destmergebranch(repo, action=action, sourceset=sourceset,
onheadcheck=onheadcheck, destspace=destspace)
return repo[node].rev()
def desthistedit(ui, repo):
"""Default base revision to edit for `hg histedit`."""
default = ui.config('histedit', 'defaultrev')
if default is None:
revs = stack.getstack(repo)
elif default:
revs = scmutil.revrange(repo, [default])
if revs:
# The revset supplied by the user may not be in ascending order nor
# take the first revision. So do this manually.
revs.sort()
return revs.first()
return None
def stackbase(ui, repo):
revs = stack.getstack(repo)
return revs.first() if revs else None
def _statusotherbook(ui, repo):
bmheads = bookmarks.headsforactive(repo)
curhead = repo[repo._activebookmark].node()
if repo.revs('%n and parents()', curhead):
# we are on the active bookmark
bmheads = [b for b in bmheads if curhead != b]
if bmheads:
msg = _('%i other divergent bookmarks for "%s"\n')
ui.status(msg % (len(bmheads), repo._activebookmark))
def _statusotherbranchheads(ui, repo):
currentbranch = repo.dirstate.branch()
allheads = repo.branchheads(currentbranch, closed=True)
heads = repo.branchheads(currentbranch)
if repo.revs('%ln and parents()', allheads):
# we are on a head, even though it might be closed
#
# on closed otherheads
# ========= ==========
# o 0 all heads for current branch are closed
# N only descendant branch heads are closed
# x 0 there is only one non-closed branch head
# N there are some non-closed branch heads
# ========= ==========
otherheads = repo.revs('%ln - parents()', heads)
if repo['.'].closesbranch():
ui.warn(_('no open descendant heads on branch "%s", '
'updating to a closed head\n') %
(currentbranch))
if otherheads:
ui.warn(_("(committing will reopen the head, "
"use 'hg heads .' to see %i other heads)\n") %
(len(otherheads)))
else:
ui.warn(_('(committing will reopen branch "%s")\n') %
(currentbranch))
elif otherheads:
curhead = repo['.']
ui.status(_('updated to "%s: %s"\n') % (curhead,
curhead.description().split('\n')[0]))
ui.status(_('%i other heads for branch "%s"\n') %
(len(otherheads), currentbranch))
def statusotherdests(ui, repo):
"""Print message about other head"""
# XXX we should probably include a hint:
# - about what to do
# - how to see such heads
if repo._activebookmark:
_statusotherbook(ui, repo)
else:
_statusotherbranchheads(ui, repo)