On Fri, Feb 18, 2005 at 06:37:35AM +0100, Andrea Arcangeli wrote:

I'll post an updated patch suitable for merging with the high-performance code as soon as I finish to port it and test it.

here we go, I improved it even further, now it handles 225-230req/sec vs 180req/sec with still full html caching with the cleaner but significantly slower ICache adapter. I found that not doing any deferred stuff at all, even if it risks to call multiple rendering paysoff measurably since the fast path won't have to execute a number of checks. I doubt it can get any faster than 230req/sec without moving into twisted space. But 230 are more than enough ;), not comparable to the ~5req/sec I get w/o caching. This below is the diff against the caching branch (and it should really be applied at least to the caching branch since there's no way the ICache on top of tags.cached like in current caching branch can reach the same performance I get with the below). Note the API is absolutely unchanged, I had not to change a line of my app after I ported it to the caching branch the first time. Don't get me wrong the ICache is nice, but the only point of the cache is to go _fast_, the nicer (and slower) stuff already happens during the real rendering. Still it's fine to use ICache in the tags.cached since tags.cached has somewhat lower performance priority and it remains a very significant optimization. --- Nevow/nevow/rend.py.~1~ 2005-02-18 04:39:54.208615490 +0100 +++ Nevow/nevow/rend.py 2005-02-18 07:13:57.082073549 +0100 @@ -30,6 +30,7 @@ from nevow import tags from nevow import flat from nevow.util import log from nevow import util +from nevow import url import formless from formless import iformless @@ -375,6 +376,8 @@ class ChildLookupMixin(FreeformChildMixi self.children[name] = child +_CACHE = {} + class Page(Fragment, ConfigurableFactory, ChildLookupMixin): """A page is the main Nevow resource and renders a document loaded via the document factory (docFactory). @@ -387,11 +390,38 @@ class Page(Fragment, ConfigurableFactory beforeRender = None afterRender = None addSlash = None + cache = False lifetime = -1 + __lastCacheRendering = 0 flattenFactory = flat.flattenFactory + def hasCache(self, ctx): + if not self.cache: + return + + _now = now() # run gettimeofday only once + timeout = _now > self.__lastCacheRendering + self.lifetime and \ + self.lifetime >= 0 + c = self.lookupCache(ctx) + if timeout or c is None: + # stop other renders + self.__lastCacheRendering = _now + c = None + return c + def cacheRendered(self, ctx, c): + if not self.cache: + return + # overwrite the deferred with the data + self.storeCache(ctx, c) + def cacheIDX(self, ctx): + return str(url.URL.fromContext(ctx)) + def storeCache(self, ctx, c): + _CACHE[self.cacheIDX(ctx)] = c + def lookupCache(self, ctx): + return _CACHE.get(self.cacheIDX(ctx)) + def renderHTTP(self, ctx): ## XXX request is really ctx now, change the name here request = inevow.IRequest(ctx) @@ -413,11 +443,18 @@ class Page(Fragment, ConfigurableFactory if self.afterRender is not None: self.afterRender(ctx) - if self.buffered: + c = self.hasCache(ctx) + if c is not None: + finishRequest() + return c + + if self.buffered or self.cache: io = StringIO() writer = io.write - def finisher(result): - request.write(io.getvalue()) + def finisher(result): + c = io.getvalue() + self.cacheRendered(ctx, c) + request.write(c) finishRequest() return result else: @@ -427,9 +464,6 @@ class Page(Fragment, ConfigurableFactory return result doc = self.docFactory.load() - if self.cache: - name = url.URL.fromContext(ctx).path - doc = tags.cached(name, self.lifetime)[doc] ctx = WovenContext(ctx, tags.invisible[doc]) return self.flattenFactory(doc, ctx, writer, finisher) @@ -504,7 +538,6 @@ class Page(Fragment, ConfigurableFactory else: ## Use the redirectAfterPost url ref = str(redirectAfterPost) - from nevow import url refpath = url.URL.fromString(ref) magicCookie = str(now()) refpath = refpath.replace('_nevow_carryover_', magicCookie) And this below is the patch of caching branch from dialtone + my above incremental patch that speedup the rend.Page caching without changing the API at all, all against trunk (i.e. the code that's going online in a few more minutes ;). Note that the original bugreport I sent in this thread is still unsolved, I've no idea why the current caching branch doesn't work in rend.Page, and I recommend dialtone to look into it, just in case the same bug can happen even while using tags.cached normally (and I use tags.cached normally too to cache some component). I think the below might be good enough for merging into trunk too. Index: Nevow/nevow/tags.py =================================================================== --- Nevow/nevow/tags.py (revision 1228) +++ Nevow/nevow/tags.py (working copy) @@ -25,7 +25,7 @@ """ -from nevow.stan import Proto, Tag, directive, raw, xml, CommentProto, invisible, slot, cdata +from nevow.stan import Proto, Tag, directive, raw, xml, CommentProto, invisible, slot, cdata, cached comment = CommentProto() @@ -62,7 +62,9 @@ def inlineJS(s): return script(type="text/javascript", language="JavaScript")[xml('\n//<![CDATA[\n%s\n//]]>\n' % s)] -__all__ = tags + ['invisible', 'comment', '_dir', '_del', '_object', '_map', 'drange', 'Tag', 'directive', 'xml', 'raw', 'slot', 'cdata', 'inlineJS'] + ['_%s' % x for x in range(100)] +__all__ = tags + ['invisible', 'comment', '_dir', '_del', '_object', + '_map', 'drange', 'Tag', 'directive', 'xml', 'raw', + 'slot', 'cached', 'cdata', 'inlineJS'] + ['_%s' % x for x in range(100)] ######################## Index: Nevow/nevow/__init__.py =================================================================== --- Nevow/nevow/__init__.py (revision 1228) +++ Nevow/nevow/__init__.py (working copy) @@ -138,6 +138,8 @@ nevow.util.remainingSegmentsFactory nevow.context.RequestContext nevow.inevow.IRemainingSegments nevow.util.currentSegmentsFactory nevow.context.RequestContext nevow.inevow.ICurrentSegments +nevow.cache.SiteCache nevow.context.SiteContext nevow.inevow.ICache + nevow.query.QueryContext nevow.context.WovenContext nevow.inevow.IQ nevow.query.QueryLoader nevow.inevow.IDocFactory nevow.inevow.IQ nevow.query.QueryList __builtin__.list nevow.inevow.IQ @@ -186,6 +188,7 @@ nevow.flat.flatstan.RendererSerializer nevow.inevow.IRenderer nevow.flat.flatstan.DirectiveSerializer nevow.stan.directive nevow.flat.flatstan.SlotSerializer nevow.stan.slot +nevow.flat.flatstan.CachedSerializer nevow.stan.cached nevow.flat.flatstan.ContextSerializer nevow.context.WovenContext nevow.flat.flatstan.DeferredSerializer twisted.internet.defer.Deferred nevow.flat.flatstan.DeferredSerializer twisted.internet.defer.DeferredList Index: Nevow/nevow/flat/flatstan.py =================================================================== --- Nevow/nevow/flat/flatstan.py (revision 1228) +++ Nevow/nevow/flat/flatstan.py (working copy) @@ -8,11 +8,15 @@ from nevow import util from nevow.stan import Proto, Tag, xml, directive, Unset, invisible -from nevow.inevow import IRenderer, IRendererFactory, IGettable, IData -from nevow.flat import precompile, serialize +from nevow.inevow import IRenderer, IRendererFactory, IGettable, IData, ICache +from nevow.flat import precompile, serialize, iterflatten from nevow.accessors import convertToData from nevow.context import WovenContext +from time import time as now +from cStringIO import StringIO +from twisted.internet import defer + allowSingleton = ('img', 'br', 'hr', 'base', 'meta', 'link', 'param', 'area', 'input', 'col', 'basefont', 'isindex', 'frame') @@ -226,6 +230,43 @@ return serialize(original.default, context) return serialize(data, context) +def CachedSerializer(original, context): + cache = ICache(original.scope(context)) + cached = cache.get(original.key, original.lifetime) + if cached: + yield cached + return + io = StringIO() + for child in iterflatten(original.children, context, io.write, + lambda item: True): + if isinstance(child, tuple): + childDeferred, childReturner = child + + d = defer.Deferred() ## A new deferred for the outer loop, whose result + ## we don't care about, because we don't want the outer loop to write + ## anything when this deferred fires -- only when the entire for loop + ## has completed and we have all the "children" flattened + + def innerDeferredResultAvailable(result): + childReturner(result) ## Cause the inner iterflatten to continue + d.callback('') ## Cause the outer iterflatten to continue + return '' + + childDeferred.addCallback(innerDeferredResultAvailable) + + ## Make the outer loop wait on our new deferred. + ## We call the new deferred back with '' + ## Which will cause the outer loop to write '' to the request, + ## which doesn't matter. It will then call our "returner", + ## which is just the noop lambda below, because we don't care + ## about the return result of the new deferred, which is just + ## '' + + yield d, lambda result: '' + result = io.getvalue() + cache.set(result, original.key) + yield result + def ContextSerializer(original, context): originalContext = original.clone(deep=False) originalContext.precompile = context and context.precompile or False Index: Nevow/nevow/stan.py =================================================================== --- Nevow/nevow/stan.py (revision 1228) +++ Nevow/nevow/stan.py (working copy) @@ -119,8 +119,40 @@ """ raise NotImplementedError, "Stan slot instances are not iterable." +def passThrough(_): + return _ +class cached(object): + """Marker for cached content + """ + __slots__ = ['key', 'children', 'lifetime', 'scope'] + def __init__(self, key, scope=None, lifetime=-1): + self.key = key + self.children = [] + self.lifetime = lifetime + self.scope = scope + if not scope: + self.scope = passThrough + + + def __repr__(self): + return "cached('%s','%s')" % self.key, self.lifetime + + def __getitem__(self, children): + """cached content is what is being cached + """ + if not isinstance(children, (list, tuple)): + children = [children] + self.children.extend(children) + return self + + def __iter__(self): + """Prevent an infinite loop if someone tries to do + for x in cached('foo'): + """ + raise NotImplementedError, "Stan slot instances are not iterable." + class Tag(object): """Tag instances represent XML tags with a tag name, attributes, and children. Tag instances can be constructed using the Prototype Index: Nevow/nevow/inevow.py =================================================================== --- Nevow/nevow/inevow.py (revision 1228) +++ Nevow/nevow/inevow.py (working copy) @@ -98,8 +98,24 @@ ANY python object is said to implement IData. """ +class ICache(compy.Interface): + """This object represents the cache that contains all the + pre-flattened fragments + """ + def get(self, index, lifetime): + """ Get an object from the cache with the given index only if + it is less old than lifetime, otherwise return None. + """ + def set(self, toBeCached, *indexes): + """ Register toBeCached with each of the indexes passed """ + def clear(self, what): + """ Clear what keyed element from the cache, or search for + what in sequences in all the keys and clear the item + """ + + class IGettable(compy.Interface): def get(self, context): """Return the data Index: Nevow/nevow/rend.py =================================================================== --- Nevow/nevow/rend.py (revision 1228) +++ Nevow/nevow/rend.py (working copy) @@ -30,6 +30,7 @@ from nevow import flat from nevow.util import log from nevow import util +from nevow import url import formless from formless import iformless @@ -375,6 +376,8 @@ self.children[name] = child +_CACHE = {} + class Page(Fragment, ConfigurableFactory, ChildLookupMixin): """A page is the main Nevow resource and renders a document loaded via the document factory (docFactory). @@ -388,8 +391,37 @@ afterRender = None addSlash = None + cache = False + lifetime = -1 + __lastCacheRendering = 0 + flattenFactory = flat.flattenFactory + def hasCache(self, ctx): + if not self.cache: + return + + _now = now() # run gettimeofday only once + timeout = _now > self.__lastCacheRendering + self.lifetime and \ + self.lifetime >= 0 + c = self.lookupCache(ctx) + if timeout or c is None: + # stop other renders + self.__lastCacheRendering = _now + c = None + return c + def cacheRendered(self, ctx, c): + if not self.cache: + return + # overwrite the deferred with the data + self.storeCache(ctx, c) + def cacheIDX(self, ctx): + return str(url.URL.fromContext(ctx)) + def storeCache(self, ctx, c): + _CACHE[self.cacheIDX(ctx)] = c + def lookupCache(self, ctx): + return _CACHE.get(self.cacheIDX(ctx)) + def renderHTTP(self, ctx): ## XXX request is really ctx now, change the name here request = inevow.IRequest(ctx) @@ -411,11 +443,18 @@ if self.afterRender is not None: self.afterRender(ctx) - if self.buffered: + c = self.hasCache(ctx) + if c is not None: + finishRequest() + return c + + if self.buffered or self.cache: io = StringIO() writer = io.write def finisher(result): - request.write(io.getvalue()) + c = io.getvalue() + self.cacheRendered(ctx, c) + request.write(c) finishRequest() return result else: @@ -499,7 +538,6 @@ else: ## Use the redirectAfterPost url ref = str(redirectAfterPost) - from nevow import url refpath = url.URL.fromString(ref) magicCookie = str(now()) refpath = refpath.replace('_nevow_carryover_', magicCookie) Index: Nevow/nevow/guard.py =================================================================== --- Nevow/nevow/guard.py (revision 1228) +++ Nevow/nevow/guard.py (working copy) @@ -24,7 +24,7 @@ from twisted.protocols import http # Nevow imports -from nevow import inevow, url, stan +from nevow import inevow, url, stan, cache def _sessionCookie(): @@ -315,6 +315,7 @@ path="/%s" % '/'.join(request.prepath), secure=secure) sz = self.sessions[newCookie] = self.sessionFactory(self, newCookie) + sz.setComponent(inevow.ICache, cache.SessionCache()) sz.args = request.args sz.fields = getattr(request, 'fields', {}) sz.content = request.content As usual this below nosense, no idea why it fixed things for me... Index: Nevow/nevow/vhost.py =================================================================== --- Nevow/nevow/vhost.py (revision 1228) +++ Nevow/nevow/vhost.py (working copy) @@ -19,7 +19,7 @@ """ def getStyleSheet(self): - return self.stylesheet + return VirtualHostList.stylesheet def data_hostlist(self, context, data): return self.nvh.hosts.keys()