[chromium-dev] Re: Chrome cache system
M. Ricardo thanks for your hints I still don't know what is the index file used for if it's not filled ! I read index is a hash table that map all cache enties... confusing 2008/10/6 Ricardo Vargas [EMAIL PROTECTED]: The cache stores whatever comes over the wire, so if the server is using compression, the data will be compressed. From your page, the following resources are cached: http://rafb.net/nopaste_favicon.gif http://rafb.net/styles/nopaste_print.css http://rafb.net/styles/nopaste.css http://rafb.net/p/S3NMeh91.html However, your code is attempting to read the entries right after the index header (from the same file), and that is not the case. The entries themselves are stored on data_xx files (you'll have to read from data_1 to see EntryStore structures). You can see code that reads an entry from disk on BackendImpl::NewEntry(), (net/disk_cache/backend_impl.cc). After that, EntryImpl::ReadData() (net/disk_cache/entry_impl.cc) is used to read the two chunks of data that you see when you use view_cache. See net/disk_cache/addr.h to understand what a cache address means. On Mon, Oct 6, 2008 at 5:23 AM, Mils Dz [EMAIL PROTECTED] wrote: Hi back. Well guys there's something I don't understand ! I initialized a cache, and then visited a simple web page, 3 entries has been added to that cache (tested with cache viewer). Now I'm supposed to find 1 entry referencing the page in the index file but I'm getting nothing ôO When I read the index header I find 3 for the num_entries field, but I only find 0 values... here's my test app: http://rafb.net/p/S3NMeh91.html Am I missing something ? Thank you 2008/10/4 Mils Dz [EMAIL PROTECTED]: A didn't thought about this. Thank you :-) And some hints about funtions used to write down cache info? Thanks :) 2008/10/4 Den Molib [EMAIL PROTECTED]: Mils Dz wrote: Guys. I'm very sorry I reread the comments and now I know why it creates those f_XXX files ...etc But still confused about the Bin/txt modes. Cheers Maybe they're stored compressed if compression was used by the server to transmit? -- www.auresdev.org Free and open source project -- www.auresdev.org Free and open source project -- www.auresdev.org Free and open source project --~--~-~--~~~---~--~~ You received this message because you are subscribed to the Google Groups Chromium-dev group. To post to this group, send email to chromium-dev@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/chromium-dev?hl=en -~--~~~~--~~--~--~---
[chromium-dev] RefPtr
If you're hacking on WebKit and don't feel like you fully understand RefPtr/PassRefPtr, you should read this: http://webkit.org/coding/RefPtr.html Darin (Adler) just updated it last night. I think it's a useful intro into ref-counting in WebCore. -eric --~--~-~--~~~---~--~~ You received this message because you are subscribed to the Google Groups Chromium-dev group. To post to this group, send email to chromium-dev@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/chromium-dev?hl=en -~--~~~~--~~--~--~---
[chromium-dev] Re: Chrome cache system
// The structure of the whole index file.
struct Index {
IndexHeader header;
CacheAddr table[kIndexTablesize]; // Default size. Actual size
controlled
// by header.table_len.
};
A given entry on the cache is identified by its key (the name of the
resource). A 32-bit hash is generated from the key. The low order bits of
the hash are used to index the actual table (usually 16 bits), and the value
stored on the table is just a cache address that indicates if the entry is
actually part of the cache, and where to find it. The entry itself is not
located on the index file.
The function that I mention before performs the last part of the lookup:
once you have the CacheAddr for the entry that you're looking for, it reads
the entry from the appropriate place. The external interface of the cache is
located on net/disk_cache/disk_cache.h, and it is implemented on
net/disk_cache/backend_impl.cc. The basic methods to locate an entry
are OpenEntry and CreateEntry, but of course they assume that you know what
you are looking for (require a key). Through them you can see how the index
file is used.
If you don't know what you're looking for, OpenNextEntry is the method to
use, but following that code is harder to see what the index file is doing.
On Mon, Oct 6, 2008 at 11:48 PM, Mils Dz [EMAIL PROTECTED] wrote:
M. Ricardo thanks for your hints
I still don't know what is the index file used for if it's not filled !
I read index is a hash table that map all cache enties... confusing
2008/10/6 Ricardo Vargas [EMAIL PROTECTED]:
The cache stores whatever comes over the wire, so if the server is using
compression, the data will be compressed.
From your page, the following resources are cached:
http://rafb.net/nopaste_favicon.gif
http://rafb.net/styles/nopaste_print.css
http://rafb.net/styles/nopaste.css
http://rafb.net/p/S3NMeh91.html
However, your code is attempting to read the entries right after the
index
header (from the same file), and that is not the case. The entries
themselves are stored on data_xx files (you'll have to read from data_1
to
see EntryStore structures).
You can see code that reads an entry from disk on
BackendImpl::NewEntry(),
(net/disk_cache/backend_impl.cc). After that, EntryImpl::ReadData()
(net/disk_cache/entry_impl.cc) is used to read the two chunks of data
that
you see when you use view_cache. See net/disk_cache/addr.h to understand
what a cache address means.
On Mon, Oct 6, 2008 at 5:23 AM, Mils Dz [EMAIL PROTECTED] wrote:
Hi back.
Well guys there's something I don't understand !
I initialized a cache, and then visited a simple web page, 3 entries
has been added to that cache (tested with cache viewer).
Now I'm supposed to find 1 entry referencing the page in the index
file but I'm getting nothing ôO
When I read the index header I find 3 for the num_entries field,
but I only find 0 values...
here's my test app: http://rafb.net/p/S3NMeh91.html
Am I missing something ?
Thank you
2008/10/4 Mils Dz [EMAIL PROTECTED]:
A didn't thought about this.
Thank you :-)
And some hints about funtions used to write down cache info?
Thanks :)
2008/10/4 Den Molib [EMAIL PROTECTED]:
Mils Dz wrote:
Guys. I'm very sorry I reread the comments and now I know why it
creates those f_XXX files ...etc
But still confused about the Bin/txt modes.
Cheers
Maybe they're stored compressed if compression was used by the server
to
transmit?
--
www.auresdev.org
Free and open source project
--
www.auresdev.org
Free and open source project
--
www.auresdev.org
Free and open source project
--~--~-~--~~~---~--~~
You received this message because you are subscribed to the Google Groups
Chromium-dev group.
To post to this group, send email to chromium-dev@googlegroups.com
To unsubscribe from this group, send email to [EMAIL PROTECTED]
For more options, visit this group at
http://groups.google.com/group/chromium-dev?hl=en
-~--~~~~--~~--~--~---
[chromium-dev] Re: Chrome Bookmarks/Passcards
Hi Chris, Scott Violet is probably going to begin doing work on a new Bookmark Manager in the next few weeks. We'll have mocks soon. You should look at this work and see if it meets your needs. -Ben On Mon, Oct 6, 2008 at 4:38 PM, Chris G [EMAIL PROTECTED] wrote: Hi, I really like Chrome, but it is obvious that effort has been put into compatibility with existing sites, the rendering and the multi-process secure architechture. The omnibox and opening page are great. Bookmarks and form filling have been neglected. This is what I know about where we are at the moment Without the 'always show bookmark bar' setting, the only way to load from a bookmark is via typing in the Omnibox or a new tab. There is no window that allows proper management of bookmarks. You can only select a bookmark in the Other bookmarks menu and right-click and press edit to edit that one bookmark. Bookmarks can be moved around the hierarchy by dragging and dropping. What I propose is a new Navigation menu that is accessed next to the back and forward arrows. This allows easy access to all bookmarks. Additionally it will goto the smart homepage, has a submenu to access all the hidden chrome stats pages and at the moment I have another submenu for managing the 'Chrome Memory' which I have identified as containing History, Downloads (management pages already exist for these two) Bookmarks, Passwords, Search places and Cookies. I have this coded but can't find a way to upload a jpg. The backend storage for history has been improved already, but the others have been ignored. I would propose looking into a way to merge all these. I have lots of ideas for managing bookmarks and integrating this with the Password system. I use RoboForm for passwords and form filling in IE and miss it in Chrome so I am happy to write the equivalent code to add something better to chrome. I would like to add many properties to each bookmark, while keeping the interface very clean and simple: * Organise bookmarks with folders and labels. * Pin bookmark to smart homepage. * Exclude bookmark from smart homepage. * Set bookmark for automatic incognito. * Mark bookmark to show on bookmark bar. * Mark bookmark to show on Navigation menu. * Store passwords and other form entries with bookmark for auto filling. Need password to use passcards and another to view/edit them. * Store favicon and thumbnail for all bookmarked sites. (anyone know why favicons are only stored in history?) * Live access to bookmark stores of other browsers and web services. * Tie bookmarks to SSL entity for better user security. * Report visit history and cookies per bookmark. * Automatically add Search features from favourite sites. * Set site to prefetch all links on page when loading for fast or offline access (Don't clog the pipes! hehe) * Implement more ways to access bookmarks using the smart home page as a start for the idea. I am asking what else is being done so I don't spend weeks writing it to find someone has uploaded something similar!! Chris PS. Pls let me know how to send you all my nav menu jpg. I also got to the bottom of a problem when adding icons to menu items changed the whole menu appearance in Vista! There are two separate classes for 'owner drawing' menus. The page and settings menu use the wrong one! --~--~-~--~~~---~--~~ You received this message because you are subscribed to the Google Groups Chromium-dev group. To post to this group, send email to chromium-dev@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/chromium-dev?hl=en -~--~~~~--~~--~--~---
[chromium-dev] Re: webkit unforking
To me, it looks like the case where we could create an abstraction, similar to what I did with ExceptionContext, which encapsulates ExecState* and ArgList* on JSC side and provides a way to pass strings/line numbers for V8. So, at this point I would merge Console.cpp into one two-chunked file (option 1) with the plan to implement and upstream ConsoleContext. ... Perhaps ExceptionContext ConsoleContext = ScriptContext? :DG On Tue, Oct 7, 2008 at 2:27 PM, Elliot Glaysher [EMAIL PROTECTED] wrote: As the person who threw together a good chunk of Console.cpp, I'd prefer that we keep Console.cpp local, at least in the short term because it's nowhere near complete yet. I'm currently waiting for crbug.com/2960 to be fixed in V8 so Console output has line numbers, which will require me to change the declarations in the USE(V8) section of Console.h so it also takes a line number... -- Elliot On Tue, Oct 7, 2008 at 2:17 PM, Ojan Vafai [EMAIL PROTECTED] wrote: If you are helping with the next merge, or with unforking WebKit code, read on. Otherwise, don't bother. In attempting to get the KJS build compiling I ran into a case that seems similar to many of the issues we've run into that required us to move things into port. I figured we could talk now about what the right solution is both so I can fix the error in front of me and so we'll have a good sense of how to move forward on these issues. In Console.h, most of the methods take KJS specific types and we fork them to take Strings, e.g. #if USE(JSC) void debug(KJS::ExecState*, const KJS::ArgList); #elif USE(V8) void debug(const String message); #endif Then we have our own implementation of Console.cpp that implements just the V8 methods and is a totally different implementation than the KJS equivalents. I see a couple solutions: 1. Move our V8 implementation of Console.cpp methods into the third_party/WebKit one wrapped in #if USE(V8). 2. Move Console.cpp out of port.vcproj and into V8Bindings.vcproj and KJSBindings.vcproj appropriatly. Are there other options? Preferences? Ojan --~--~-~--~~~---~--~~ You received this message because you are subscribed to the Google Groups Chromium-dev group. To post to this group, send email to chromium-dev@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/chromium-dev?hl=en -~--~~~~--~~--~--~---
[chromium-dev] Re: webkit unforking
Eric asked on #webkit, and (since I don't think they log(?)) I'll just paraphrase a response from there, which is that this is a bad idea since the inspector needs to have all the arguments and not just a single string. On Tue, Oct 7, 2008 at 2:17 PM, Ojan Vafai [EMAIL PROTECTED] wrote: If you are helping with the next merge, or with unforking WebKit code, read on. Otherwise, don't bother. In attempting to get the KJS build compiling I ran into a case that seems similar to many of the issues we've run into that required us to move things into port. I figured we could talk now about what the right solution is both so I can fix the error in front of me and so we'll have a good sense of how to move forward on these issues. In Console.h, most of the methods take KJS specific types and we fork them to take Strings, e.g. #if USE(JSC) void debug(KJS::ExecState*, const KJS::ArgList); #elif USE(V8) void debug(const String message); #endif Then we have our own implementation of Console.cpp that implements just the V8 methods and is a totally different implementation than the KJS equivalents. I see a couple solutions: 1. Move our V8 implementation of Console.cpp methods into the third_party/WebKit one wrapped in #if USE(V8). 2. Move Console.cpp out of port.vcproj and into V8Bindings.vcproj and KJSBindings.vcproj appropriatly. Are there other options? Preferences? Ojan --~--~-~--~~~---~--~~ You received this message because you are subscribed to the Google Groups Chromium-dev group. To post to this group, send email to chromium-dev@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/chromium-dev?hl=en -~--~~~~--~~--~--~---
[chromium-dev] Re: webkit unforking
There are a bajillion logs of #webkit. None of them official. -eric On Tue, Oct 7, 2008 at 2:50 PM, Evan Martin [EMAIL PROTECTED] wrote: Eric asked on #webkit, and (since I don't think they log(?)) I'll just paraphrase a response from there, which is that this is a bad idea since the inspector needs to have all the arguments and not just a single string. On Tue, Oct 7, 2008 at 2:17 PM, Ojan Vafai [EMAIL PROTECTED] wrote: If you are helping with the next merge, or with unforking WebKit code, read on. Otherwise, don't bother. In attempting to get the KJS build compiling I ran into a case that seems similar to many of the issues we've run into that required us to move things into port. I figured we could talk now about what the right solution is both so I can fix the error in front of me and so we'll have a good sense of how to move forward on these issues. In Console.h, most of the methods take KJS specific types and we fork them to take Strings, e.g. #if USE(JSC) void debug(KJS::ExecState*, const KJS::ArgList); #elif USE(V8) void debug(const String message); #endif Then we have our own implementation of Console.cpp that implements just the V8 methods and is a totally different implementation than the KJS equivalents. I see a couple solutions: 1. Move our V8 implementation of Console.cpp methods into the third_party/WebKit one wrapped in #if USE(V8). 2. Move Console.cpp out of port.vcproj and into V8Bindings.vcproj and KJSBindings.vcproj appropriatly. Are there other options? Preferences? Ojan --~--~-~--~~~---~--~~ You received this message because you are subscribed to the Google Groups Chromium-dev group. To post to this group, send email to chromium-dev@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/chromium-dev?hl=en -~--~~~~--~~--~--~---
[chromium-dev] Re: Issue 3080: {Get/Set}CurrentDirectory()
Their fixed-size nature is also pretty lousy. That is to say, they're
limited to MAX_PATH with no provision for long-filenames (longer than
MAX_PATH that is). They're not the only place that Win32 has such
restrictions, but they're one of the harder to avoid ones.
2008/10/6 Darin Fisher [EMAIL PROTECTED]:
I added a comment in the bug, which explains why I think those functions are
to be avoided. There may be other good reasons too!
-Darin
On Sun, Oct 5, 2008 at 8:27 PM, enchantingBlu [EMAIL PROTECTED]
wrote:
Hi,
A bug has been filed lately for replacing GetCurrentDirectory and
SetCurrentDirectory. No furthur information is provided. I assume the
problem with these would be about DOS kind of naming and full path
naming of the directory paths? Also, with the cross platform work
going on, these will be more or less consumed anyway. But if anyone
has any real issues with these then lets discuss this here or get the
bug out of the backlog?
thanks,
Zaki
--
char a[9],*p=a;main(c,V)char**V;{char*v=c0?1[V]:V;if(c)for(;(c=*v)93^
c;p+=!(62^c)-!(60^c),*p+=!(43^c)-!(45^c),44^c||read(0,p,1),46^c||putchar(*p)
,91^c||(v=*p?main(-1,v+1),v-1:main(0,v)),++v);else for(;c+=!(91^*v)-!(93^*v)
;++v);return v;} /* [EMAIL PROTECTED]brainf*** program as argv[1] */
--~--~-~--~~~---~--~~
You received this message because you are subscribed to the Google Groups
Chromium-dev group.
To post to this group, send email to chromium-dev@googlegroups.com
To unsubscribe from this group, send email to [EMAIL PROTECTED]
For more options, visit this group at
http://groups.google.com/group/chromium-dev?hl=en
-~--~~~~--~~--~--~---
