Jump to content

mixit

Member
  • Posts

    222
  • Joined

  • Days Won

    10
  • Donations

    0.00 USD 

Posts posted by mixit

  1. FWIW, this 360 memory usage issue doesn't seem to depend (much) on hardware or 32/64-bit XP.

    This is how it behaves on two of my systems (deliberately not giving their specs).
    The columns are Private bytes/Peak private bytes/Virtual size/Peak virtual size per Process Hacker

    #1 XP x64 safe mode:
     40.78 MB,  43.53 MB, 298.88 MB, 320.48 MB
     22.66 MB,  22.73 MB, 237.16 MB, 240.91 MB
     12    MB,   12.9 MB, 212.93 MB, 215.56 MB
     14.72 MB,  15.12 MB, 219.1  MB, 220.35 MB
     18.53 MB,  20.85 MB, 233.04 MB, 233.04 MB
     27.84 MB,  35.43 MB, 243.05 MB, 252.11 MB
    -----------------------------------------
    136.53 MB            1444.16 MB
    
    #1 XP x64 normal:
    144.02 MB, 149.41 MB, 337.77 MB, 359.34 MB
    124.61 MB, 124.71 MB, 247.54 MB, 255.79 MB
    118.21 MB, 120.66 MB, 240.08 MB, 243.83 MB
    116.56 MB, 119.95 MB, 222.2  MB, 223.45 MB
    121.02 MB, 123.12 MB, 233.58 MB, 234.83 MB
    130.8  MB, 140.68 MB, 248.53 MB, 255.81 MB
    -----------------------------------------
    755.22 MB            1529.7  MB
    
    #1 XP x64 fresh install:
    (saved partition, NO drivers/software that aren't from the official CD)
     38.84 MB,  43.11 MB, 305.51 MB, 325.8  MB
     22.96 MB,  23.04 MB, 233.99 MB, 237.77 MB
     12.43 MB,  13.38 MB, 212.94 MB, 215.88 MB
     15.06 MB,  15.47 MB, 217.15 MB, 218.4  MB
     19    MB,  21.1  MB, 229.52 MB, 229.52 MB
     28.12 MB,  34.22 MB, 252.29 MB, 261.35 MB
    -----------------------------------------
    136.41 MB            1451.4  MB
    
    
    
    #2 XP x86 safe mode:
     33.31 MB,  35.57 MB, 279.27 MB, 291.3  MB
     19.54 MB,  19.58 MB, 207.03 MB, 210.56 MB
      9.03 MB,  10.02 MB, 190.2  MB, 191.7  MB
     12.9  MB,  14.09 MB, 716.5  MB, 717.5  MB
     16.59 MB,  20.57 MB, 723.69 MB, 726.38 MB
     25.76 MB,  34.08 MB, 741.77 MB, 744.52 MB
    -----------------------------------------
    117.13 MB            2858.46 MB
    
    #2 XP x86 normal:
    137.5  MB, 140.19 MB, 290.51 MB, 310.59 MB
    121.35 MB, 121.39 MB, 211.94 MB, 217.47 MB
    113.72 MB, 118.63 MB, 207.76 MB, 209.26 MB
    115.16 MB, 118.66 MB, 708.92 MB, 709.92 MB
    119.27 MB, 122.78 MB, 729.05 MB, 731.98 MB
    127.8  MB, 135.48 MB, 741.06 MB, 745.31 MB
    -----------------------------------------
    734.8  MB            2889.24 MB

    It's interesting that virtual sizes are quite close (per OS/HW) even when RAM usage differs greatly between safe mode and "normal" boot. But clearly something has happened between the fresh install and my current configuration to cause this difference.

    Edit: The 360 profile used was exactly the same, copied from one system to the other.

  2. What would really be great is if we didn't have to jump through any loops to disable this menu stuff... I'm not talking about disparate and still limited settings like with official Chrome, but the way it used to be possible with XUL and extensions like Menu Wizard. I think I have somewhere between 2/3 and 3/4 or all menus/items turned off with it, browsing is much more convenient without all the clutter. :yes:

  3. 18 minutes ago, UCyborg said:

    Aren't you also using uBlock Origin? I haven't seen any ads with this one alone.

    OT: Seconded. I generally use uMatrix with uBO only turned on for specific sites for cosmetic or script mods, and after copying just the following from the standard uBO lists (I don't use them in their entirety, no need with uMatrix), I haven't seen any ads on Youtube (except of course when casters plug stuff as part of their actual streams).

    Quote
    ! Copied from official uBlock filters
    ! http://www.wilderssecurity.com/threads/ublock-a-lean-and-fast-blocker.365273/page-25#post-2461804
    ! Specific cosmetic filters for Youtube home page
    youtube.com##.masthead-ad-control,.ad-div,.pyv-afc-ads-container
    ! https://github.com/uBlockOrigin/uAssets/issues/644#issuecomment-324214893
    youtube.com###promotion-shelf
    ! https://github.com/uBlockOrigin/uAssets/issues/1408
    *_ad_$media,domain=youtube.com,3p
    ! https://redd.it/fg424w
    youtube.com##ytd-video-masthead-ad-advertiser-info-renderer,ytm-promoted-sparkles-web-renderer
    ! https://github.com/easylist/easylist/issues/5112
    ! @@||youtube.com/get_video_info?*timedtext_editor$xhr,1p
    ! https://redd.it/ggcmkp https://redd.it/gx03e0
    !youtube.com,youtube-nocookie.com##+js(json-prune.js, *.playerResponse.adPlacements *.playerResponse.playerAds [].playerResponse.adPlacements [].playerResponse.playerAds playerResponse.adPlacements playerResponse.playerAds adPlacements playerAds)
    youtube.com,youtube-nocookie.com##+js(json-prune.js, [].playerResponse.adPlacements [].playerResponse.playerAds playerResponse.adPlacements playerResponse.playerAds adPlacements playerAds)
    youtube.com,youtube-nocookie.com##+js(set-constant.js, ytInitialPlayerResponse.adPlacements, undefined)
    youtube.com,youtube-nocookie.com##+js(set-constant.js, playerResponse.adPlacements, undefined)
    ! https://www.reddit.com/r/uBlockOrigin/comments/kg7224/
    youtube.com##ytd-display-ad-renderer:upward(ytd-rich-item-renderer)
    

    @msfntor

  4. 3 hours ago, NotHereToPlayGames said:

    Engineering Department.  And it's an HR rule that any audio be played on headphones and never on computer/laptop speakers. [...]

    they used to block YouTube entirely, but they found it forced them to loosen restriction on USB Ports so that we could listen to music that way instead.

    OT: Don't they allow you to bring your phone or a music player to the office so work computers needn't be involved? (I walk around with my Sansa Clip even at home and use Youtube mostly to watch videos, not for plain listening.) That ad thing I could totally see, though :)

  5. 8 hours ago, rereser said:

    found no page load speed increase when browsing with the translate portion changed.

    If there was any speed up as a side effect, it would likely only be noticable with foreign language pages. But I think the logic that got dropped makes up too small a part of what's involved in page rendering. I guess what I was trying to say was that all other changes just involved jumping over showing a menu item (which definitely wouldn't be perceivable) and this was the only one that involved a bit more.

  6. On 11/26/2022 at 2:21 AM, mina7601 said:

    I'm probably doing something wrong/missing something here, but there's no actually "TranslateEnabled" registry entry at all. I tried using ProcMon as per @mixit's post, but nothing specific comes up. I tested Ungoogled Chromium 86.0.4240.193 32-bit.

    No, you're quite right, it wouldn't be there unless the policy is specifically applied, and Ungoogled has no reason to follow Google-branch policies. I thought maybe they have their own policies under a different key, but it's not too surprising if they don't.

  7. Here are the patches needed so remove some Google-related context menu items that @NotHereToPlayGames prefers not to have in his un-Googled builds. The patches are to be applied to chrome.dll from 13.5.2022.0 rebuild 2, but they aren't specific to this rebuild, so if he doesn't include all of these in his future versions, you could still apply them on your own. They should even work with the Chinese original, but not with other base builds like 1030, not to mention the 13.0 line etc.

    Fortunately everything that was needed could be derived from static analysis and the logic for these things was pretty simple, so I could get this done relatively quickly. However, this has barely been tested, so be prepared for browser crashes and things not working as they should. If in doubt, please refrain from using these patches for now. I certainly hope I got everything right the first time, but this type of work is very unforgiving of even the smallest of mistakes... It's really not my style to release something like this without real-life testing it for a while, but I wanted to get this out to @NotHereToPlayGames so now he can beta test something for me :lol:. But seriously, I don't really have time to test this myself as I'm not using this browser that much, and even then a customized older version.

    FYI, some of these changes are for "IE mode" menus. I never use the IE engine with 360 myself, but since it wasn't that hard to fix, I figured why not do it. Oh, and feel free to mix and match, the patches don't depend on each other.

    Make sure you make edits in "hex" mode, not plain text mode (not trying to patronize; I got into some weird errors when I forgot to switch to the hex tab one time). All of these edits should only match 1 time with Replace all. Red bytes are replaced with black ones, but always search/replace the entire sequence, not just the changing part.


    Drop Share URL QR Code from Chrome mode page context menu:
    89 F9 68 79 1A 00 00 68 4F 81 00 00
    EB 0F 68 79 1A 00 00 68 4F 81 00 00

    Drop Share URL QR Code from Chrome mode link context menu:
    89 F1 68 79 1A 00 00 68 50 81 00 00
    EB 0F 68 79 1A 00 00 68 50 81 00 00

    Drop Share URL QR Code from IE mode page context menu (three parts, because need to match up surrounding menu items):
    6A 00 68 D5 08 00 00 FF 75 08 FF 15 44 42 ED 16
    EB 5E 68 D5 08 00 00 FF 75 08 FF 15 44 42 ED 16

    68 E4 25 00 00 FF 75 08 FF 15 44 42 ED 16
    68 D5 08 00 00 FF 75 08 FF 15 44 42 ED 16

    68 E4 25 00 00 FF 75 08 FF 15 C4 42 ED 16
    68 D5 08 00 00 FF 75 08 FF 15 C4 42 ED 16

    Drop Share URL QR Code from IE mode link context menu:
    8D 7D E8 68 79 1A 00 00 57 E8 F3 0F 14 FB
    EB 3A 90 68 79 1A 00 00 57 E8 F3 0F 14 FB

    Drop Share URL QR Code from IE mode image link context menu:
    8D 45 D8 68 79 1A 00 00
    EB 43 90 68 79 1A 00 00

    Drop Share image location from Chrome mode image context menu:
    89 F9 68 7A 1A 00 00 68 51 81 00 00
    EB 0F 68 7A 1A 00 00 68 51 81 00 00

    Drop Share image location from IE mode image context menu:
    6A 00 6A 0F 53 FF 15 44 42 ED 16 83 F8 FF 74 3C
    EB 4A 6A 0F 53 FF 15 44 42 ED 16 83 F8 FF 74 3C

    Drop Translate to English from Chrome mode page context menu (this also avoids some of the translation-related housekeeping, could (very) slightly speed up page loads):
    8D 86 F0 02 00 00 50 E8 CF BA 71 FE
    E9 E6 00 00 00 90 50 E8 CF BA 71 FE

  8. @NotHereToPlayGames

    Not to dwell on it, but I think I should say this. I'm not sure how much of the negativity you mentioned in the old thread was attributed to me (I used the word "god", for sure), but I honestly didn't think someone with your constitution would consider abandoning the project just because I threw in that Brotli black-box issue. I actually chuckled when someone told me they were afraid of that happening. In any case, I'm sorry if I contributed to negative atmosphere and certainly hope you can continue with this fresh start for the benefit of both the members here and the wider XP circle.

  9. 1 hour ago, NotHereToPlayGames said:

    for reference (I seek to REMOVE these from the context menu completely - these "features" are intentionally broken in my default ungoogled releases).

    For the record, simply deleting or editing the labels for these doesn't seem to work, you just get blank menu items that still have the functionality. (I'm not sure that was the gist of what he was hinting at last night; if there was something else, hopefully he can elaborate.)

    1 hour ago, NotHereToPlayGames said:

    My local file had some cosmetic changes to the extensions page (personal preference) and to the settings/content page (to correct anomolies on laptops with 768px max vertical resolution) and all of these changes were within file 13570.

    Ah, OK, glad that's settled. The way you brought it up earlier seemed to suggest my paks were suspicious in some way, that's why I grumpily reacted with that line about hacking.

  10. 41 minutes ago, D.Draker said:

    On the 1st page version number 13.5.1030.0

    Well, no wonder then, because I don't have his rebuild 6 for that and both the links from the first post are 404 for me right now. An excellent excuse for bedtime. :lol:

    EDIT: But if you mean "Translate to $1" (it's slightly shifted in rebuild 5), then yes, I'm aware that's where the label is. I was talking based on https://chromium.googlesource.com/experimental/chromium/src/+/refs/tags/86.0.4240.198/chrome/browser/renderer_context_menu/render_view_context_menu.cc , or I guess https://chromium.googlesource.com/chromium/src/+/refs/tags/86.0.4240.198/chrome/browser/renderer_context_menu/render_view_context_menu.cc is a more proper link (the content is exactly the same).

     

  11. 43 minutes ago, D.Draker said:

    I'm sorry, you both look in the wrong places . I shall give you a hint, it's at offset 10510 in the English locale pak file. But I'm on version 13.5. And it looks like @NotHereToPlayGames is not interested in me being be here, "'cause your're on Vista" crap. So I'm helping only out of respect to @Dave-H.

    No you're not :lol: (sorry, I mean). But I'm really going to sleep now, 36 hrs uptime and hints about "the" god knows which build version don't enlighten me right now. :zzz:

  12. @NotHereToPlayGames Geez, why are you wasting precious holiday time on what's not very interesting for you when you could just give me that resource.pak of yours so I could compare it with both my modded version and your previously released 2022? :lol: No Thanksgiving here, though I probably will have to go to sleep soon, been up for more than 24 hrs (not related to my neck pain, that's pretty much OK now).

    EDIT: Regarding your two context menu items, unfortunately it's leaning towards having to patch the DLL code to get these to go away. Going by the original 86.0.4240.198 Chromium, the QR item should be controllable via chrome://flags/#sharing-qr-code-generator , but it seems to have no effect in 360. The translation item should be blockable via TranslateEnabled policy key in the registry, but 360 doesn't seem interested in Chrome or Chromium policies (it reads a single one from the latter, unhelpful for this). Since I don't feel strongly about these items, I can't promise I'll be spending a lot of time tracking this stuff down in the code, I guess we'll have to see...

  13. 52 minutes ago, NotHereToPlayGames said:

    I'm more interested in trying to track down where the "Translate to English" and "Share URL QR Code" context menus are coming from.  Have you been able to isolate them?

    Like I said, if you could just upload your .pak I can do it myself, this difference bothers me. As for the context menus, no, I haven't looked into them.

  14. 3 minutes ago, NotHereToPlayGames said:

    Performing these three file changes and these three only, our resulting resources.pak file are not the same size.

    I'm not overly concerned as I don't criss-cross anything on what I use for me-myself.  Mentioning it only for the sake of those here that may be criss-crossing.

    image.png.56a747bdf595def17fd662ac58e4b2a9.png

    That's because I hacked the hell out of these 16 bytes to pwn all the fools who downloaded my pak, of course. :rolleyes:

  15. 1 hour ago, NotHereToPlayGames said:

    Because of the collection not matching!  I personally trust the collection from "chrome-pak-customizer" over-and-above the collection from Python.  So I'm comparing just "why" the collections do not match.

    Can't say I get why we shouldn't trust the official tools. In any case, as far as I can tell, when there are items with identical contents (listed as "aliases" in the customizer's pak_index.ini file), the customizer extracts only the first item encountered (and marks the rest in the .ini), while pak_util.py extracts all these items as separate files with identical contents (when repaking, these identicals get folded back into a single copy of the data).

    I guess the customizer way is more convenient if you want to change an item and have that also update its aliases without additional hassle, and the other way is more convenient if you want to de-alias/change some of the items. Neither tool is "wrong", although with the customizer it seems like you have to be careful not to get your aliases out of alignment if you add or remove items (or de-alias something), because it doesn't mark them by the actual item number but rather its line position. And of course with pak_util you have to detect which ones are identical on your own, and be careful to update all of the identicals if you don't want to de-alias them.

    1 hour ago, NotHereToPlayGames said:

    These files are only the files that 7Zip could not extract from the "chrome-pak-customizer" collection.

    OK, I'm just saying not all of these are Brotli compressed.

  16. 5 hours ago, NotHereToPlayGames said:

    For anyone that may be interested, here are the 249 files that were originally Brotli-compressed.

    https://www.dropbox.com/s/nunqk0fcv7m4efi/brotli%20decompressed%20249%20files.7z?dl=1

    Unless I have my wires crossed, this collection seems to contain a few files that aren't compressed with Brotli and miss a few that are. Try running python file_types_in_pak.py from https://pastebin.com/7kU3wMHi in the folder you unpacked to with pak_util.py with the --raw option, you should get a slightly different set. (Caution: the script will rename the files with extensions added according to type.)

    For working with individual files:

    brotli_decompress_for_pak.py : https://pastebin.com/fNyi1UdK
    brotli_compress_for_pak.py : https://pastebin.com/Z2YmGPFP

    I'm curious why you need to decompress one at a time to compare, though.  Since pak_util.py can also extract --raw, I don't really see the need for that other extractor at all. The reason being that I prefer to use the same toolset that is used to build the product. YMMV.

  17. @NotHereToPlayGames
    I'm not sure any general-purpose tool will decompress these Brotli files "as-is", because they have a non-standard header added to them:
    2 "magic bytes" 0x1E, 0x9B to identify Brotli
    6 bytes for uncompressed data size
    Brotli content

    It's (reversely) similar to 7z not extracting plain zlib compressed files unless you add a gzip header to them.

    As for recompression, technically you don't even need to recompress these files at all, or you can use the alternate gzip compression that many other files use, the one you've demonstrated with 7z. There's nothing special about Brotli compression in terms of functionality (that I've encountered, at least), it's just one delivery format among others.

    Honestly, I don't see what benefit a batch file extractor could provide for you when the Python tool already extracts everything at once and you have the capability to use it. You can make it yourself, somehow chop off the first 8 bytes and then pipe into peazip or whatever that takes regular Brotli.

    For Python, I generally use Notepad++, no IDE, just command-line execution. "Works for me." I wasn't trying to say IDLE was bad, just that you wouldn't really need it for making the edits I mentioned and running the GRIT pak utility. If you're going to do more programming in Python, sure, why not.

  18. @NotHereToPlayGames

    You may not know it, but you did use me as a beta tester, as in I used your 360ChromePortable_13.5.1030_rebuild_5_ungoogled as the basis for what I'm still using myself. And if I'd found something very dangerous in those "black-box" files, rest assured that I'd have told everyone (It's basically the Dev Tools in there). So I didn't say anything because I more or less knew what was in there and there was nothing much to say about it, but I still don't quite get why you didn't say anything if you didn't know what was in there.

    I don't know why you're (apparently) taking all this as me invalidating your entire project, because I'm absolutely not doing that. It's not my goal to crucify you or anything like that, it's just that some of your claims and logic puzzle me. (Like I don't really get the point of some of the parsing in your latest post, but whatever.) There's no question about your project having been useful, why would I deny that? I in fact admire people who take on this kind of stuff because I myself tend to be too fickle to do it (especially long-term support, I get tired of interacting with users), even if I could sometimes do a better job on some of the things somebody else is doing. So I mostly just spot when I have the ability and feel the need. But in this case I learned something pretty weird in the process of doing so, and I was in a "mood", so I brought it up. No one is the villain here.

  19. Thanks everyone for your compassion and suggestions. The neck feels better today than it did yesterday, so I hope it'll stop bothering me pretty soon. :)

    @NotHereToPlayGames

    1 hour ago, NotHereToPlayGames said:

    Almost ALL of the files "inside" resources.pak extract-and-edit "inside" of 7Zip.  Regular ol' 7Zip.  No command line and no "addons".  PIcs below.

    It is certainly "curious" that THESE THREE files do NOT.

    More like TWO HUNDRED AND FIFTY-THREE (not shouting, aping :lol:) files with Brotli compression, that's almost 1/6 of resources.pak contents (253/1575)...

    Not the odds I'd feel comfortable about (well maybe if I thought there were only "THREE" :lol:)..

    Ain't no one talking about 100% safety here, nor measuring who's been around the longest or done the most (your work is much appreciated!). But if I have no idea about something, I try not to make far-reaching claims about it. So it surprises me when people do that, that's all. Is it really so difficult to just keep things on the level of "I believe it's safe enough for my own purposes (and my logging seems to support that), BUT there's this stuff in there I have no clue about, so take that into consideration."? Actually, if you'd said something like that, I (or someone else) could have told you how to open these files long ago. Or, I don't know, maybe you have mentioned it, because I haven't read everything in these 360 threads and also your, no offense, tendency to grandstand can overshadow things. In any case, this was pretty weird for me.

×
×
  • Create New...