Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]


Groups > comp.sys.mac.advocacy > #136425

Re: It is stunning when you see how badly Windows operates: indexing

From Alan <nuh-uh@nope.com>
Newsgroups comp.sys.mac.advocacy, alt.comp.os.windows-10, alt.comp.os.windows-11
Subject Re: It is stunning when you see how badly Windows operates: indexing
Date 2025-08-10 17:00 -0700
Organization A noiseless patient Spider
Message-ID <107bbqj$26r4d$1@dont-email.me> (permalink)
References <1078llp$1he7o$3@dont-email.me> <10797gb$1mdq9$1@toylet.eternal-september.org> <107adkp$1voqj$1@dont-email.me> <107ase3$22qav$1@dont-email.me> <107bb0o$27f7h$1@dont-email.me>

Cross-posted to 3 groups.

Show all headers | View raw


On 2025-08-10 16:46, Paul wrote:
> On Sun, 8/10/2025 3:37 PM, Alan wrote:
>> On 2025-08-10 08:25, Paul wrote:
>>> On Sun, 8/10/2025 12:34 AM, Mr. Man-wai Chang wrote:
>>>> On 10/8/2025 7:30 am, Alan wrote:
>>>>> I'm going some tech support for my brother today, and that's meant
>>>>> looking "under the hood" of his Windows 10 machine.
>>>>>
>>>>> And it is astounding how poorly Windows handles indexing content.
>>>>
>>>> This Indexing Service should have been *OPTIONAL*. The search function should work regardless of the indexing settings.
>>>>
>>>> Why can't I choose a slow but more reliable method that doesn't need indexes?? :)
>>>>
>>>
>>> It *does* work without an index, or at least, with a minimally-sized index.
>>> It is horribly slow if done that way (it fills in the areas missing from
>>> the index file, via brute force search). Any time the File Explorer does something,
>>> it may feel the need to look for an icon to represent the file, and the
>>> graphical overhead slows everything down.
>>>
>>> "Everything.exe" from voidtools, is about the best you can do for a
>>> filename-only solution. The file it keeps, just has filenames in it.
>>> And for NTFS partitions, it keeps the list of files up to date, using
>>> the USN Journal information.
>>>
>>> It does not matter what OS, any time the storage has
>>> a lot of files, the behavior does not scale well. it can take
>>> an hour, just to count the files in a folder like this one.
>>> "Everything.exe" is also going to need an hour for this one,
>>> as the program does a stat() on each file. Early versions of
>>> the Voidtools program, did not include file size in the file list,
>>> and the software was a lot faster then.
>>>
>>>       [Picture]
>>>
>>>        https://i.postimg.cc/Y2NWHSNk/test-volume-large-file-set.gif
>> I'm sorry, but it does matter how cleverly and well the OS is coded to do the indexing.
>>
>> The fact of the matter is that Windows's indexing is orders of magnitude slower than macOS's indexing.
>>
>> If I move 80,000 files on macOS, the OS is smart enough to know that it doesn't need to update every piece of information about those files; only that they are now in a new location.
>>
>> The very fact that a tool like "Everything.exe" has been developed for
>> Windows shows just how poorly done the built-in indexing is.
> 
> https://en.wikipedia.org/wiki/Apple_File_System
> 
>      "Despite the ubiquity of APFS volumes in today's Macs and the format's
>       2016 introduction, third-party repair utilities continue to have notable
>       limitations in supporting APFS volumes, due to Apple's delayed release of
>       complete documentation.
> 
>       According to Alsoft, the maker of DiskWarrior, Apple's 2018 release of
>       partial APFS format documentation has delayed the creation of a version of
>       DiskWarrior that can safely rebuild APFS disks.[44] Competing products,
>       including MicroMat's TechTool and Prosoft's Drive Genius, are expected
>       to increase APFS support as well."
> 
> Not with a barge pole. See ? I've been to your rodeo, and rode that pony.
> Back when I was an Apple user, you did not own a Mac without a copy of DiskWarrior.
> Physical media with a serial number on the label, was sent to you.
> 
>       [Picture]
> 
>        https://i.postimg.cc/k5fPj6hg/discwarrior.jpg

Why is ANY of that relevant?

> 
> *******
> 
> You don't have to "cp" or "copy" a file to copy it as such.

OK. So what? I wasn't copying files when Windows wanted to re-index them...

...I was just moving them.

> 
> Windows has hardlinks, which allow the data clusters to stay put,
> and adding an additional $FILENAME to the $MFT 1KB entry is sufficient
> to make a file appear in two places at once. By deleting the first
> $FILENAME, the second $FILENAME still owns the clusters.
> 
> File 479487
> \Windows\System32\shell32.dll
>      $STANDARD_INFORMATION (resident)
>      $FILE_NAME (resident)
>      $FILE_NAME (resident)
>      $DATA (nonresident)
>          logical sectors 65604464-65622255 (0x3e90b70-0x3e950ef)
>      $EA_INFORMATION (resident)
>      $EA (resident)
>      Attribute Type 0x100 $DSC (resident)
>      Attribute Type 0x100 $TXF_DATA (resident)
> 
> The -i option in Linux, includes the inode number as the first numeric field.
> And the inode number on Linux, is made equal to the Filenum. This is how
> we know we are seeing the two $FILE_NAME values. The nfi.exe utility, sorely
> needs this information added to it, as otherwise the listing via nfi.exe misses stuff.
> 
> These two files share the same clusters.  ls -algtRi
> 
> ./Windows/System32:
> total 2371265
> 479487 -rwxrwxrwx  2 mint   9108432 Jul  9 02:02 shell32.dll
> 
> ./Windows/WinSxS/amd64_microsoft-windows-shell32_31bf3856ad364e35_10.0.22621.5547_none_4c0780878d97ec32:
> total 40256
> 573225 drwxrwxrwx 1 mint 13107200 Jul 11 11:05 ..
> 454973 drwxrwxrwx 1 mint     4096 Jul  9 02:02 .
> 479487 -rwxrwxrwx 2 mint  9108432 Jul  9 02:02 shell32.dll
> 
> *******
> 
> In other words, to copy the file during a Patch Tuesday run, the OS first
> unpacked a new version of package in WinSxS. That would be the first $FILE_NAME
> created.
> 
> Then, it created a hardlink in C:\Windows\System32 for shell32.dll , and now
> one set of clusters, has two file names. You could delete the WinSxS folder
> and the copy in System32 would still exist, and so would the clusters
> 
>     65604464-65622255  (22255-4464+1)/8 = 2224 clusters
> 
> that belong to the file.
> 
> *******
> 
> Windows has various versions of features that correspond roughly
> to things seen on other OSes. The features are not exactly the same,
> but similar.
> 
> I could copy a large file, without repeating the clusters, just by using
> a hard link. As long as that hard link stayed on the same physical partition.
> 
> People here, do not copy things that way, they would use a Move done
> with the mouse, or use some utility that is smart enough to do it
> the efficient way. The danger with a Move on any platform, is
> a power fail at some microsecond while the operation is being done.
> There is usually a crevasse your file can fall into, if you use "Move".
> Having a UPS or a battery on the equipment, makes that marginally safer.
> 
> A "Move" on FAT32, might be unsafe, while a "Move" on NTFS could be made safer.
> 
> Copying the dumb way, is what happens when you're in a rush. And when
> dealing with things like VM containers, you're not usually in that
> kind of a rush. Like, if I was moving a 2TB MRIMG backup file around,
> I would pause and think about what I was doing. Doing copy-pasta would be
> furthest from mind, because of how long it could take if I did it the
> dumbest way possible.
> 

I don't know why you've posted all of this except to deflect.

> *******
> 
> While the nfi.exe utility displays items this way...
> 
>     File 479487
>     \Windows\System32\shell32.dll
> 
> they are not stored that way. The string "shell32.dll" is in the
> 1KB entry. And an integer number like "12345" references the parent
> directory which is "System32". To print out the absolute path,
> involves walking the tree from parent to parent, until you've
> assembled all the parent names. When you reach filenum 5 at the top
> of the tree, its parent is filenum 5 (a file system loop) and
> that is how your software knows it is at the top.
> 
>     File 5            <=== there is no path field on this printout (this file has a parent of "5")
>     Root Directory
>         $STANDARD_INFORMATION (resident)
> 
> The drive letter is not stored on the partition.
> If you multiboot (have a W10 and a W11 stored on
> the same SSD), then the drive lettering is stored
> in each Registry, and a drive that is E: when one OS
> is booted, could be F: when the other OS is booted.
> 
> I'm surprised you didn't notice the worst feature of the filesystem.
> The time it takes to delete a file, has increased within the last
> two years. And I have no technical note to share with you, as to why
> this change was made, or, what they're doing which is so slow. Just
> be aware, that in addition to the excessive calculations to "enable
> a graphical animation", additional time is wasted on something we cannot
> see. Even holding down the shift key when deleting, does not make
> that time waster go faster. If it was part of the "Robust NTFS" program,
> they would gleefully have declared that.
Again: why the HELL do I care about deleting files.

Why can't you just admit the absurdity of a simple move of a file 
triggering a lengthy re-indexing process?

Back to comp.sys.mac.advocacy | Previous | NextPrevious in thread | Next in thread | Find similar | Unroll thread


Thread

It is stunning when you see how badly Windows operates: indexing Alan <nuh-uh@nope.com> - 2025-08-09 16:30 -0700
  Re: It is stunning when you see how badly Windows operates: indexing Hank Rogers <Hank@nospam.invalid> - 2025-08-09 18:55 -0500
    Re: It is stunning when you see how badly Windows operates: indexing Alan <nuh-uh@nope.com> - 2025-08-09 17:08 -0700
      Re: It is stunning when you see how badly Windows operates: indexing Hank Rogers <invalid@nospam.com> - 2025-08-10 03:22 +0000
        Re: It is stunning when you see how badly Windows operates: indexing Paul <nospam@needed.invalid> - 2025-08-09 23:52 -0400
          Re: It is stunning when you see how badly Windows operates: indexing "David B." <BD@hotmail.co.uk> - 2025-08-10 09:54 +0100
  Re: It is stunning when you see how badly Windows operates: indexing Paul <nospam@needed.invalid> - 2025-08-09 23:28 -0400
  Re: It is stunning when you see how badly Windows operates: indexing "Mr. Man-wai Chang" <toylet.toylet@gmail.com> - 2025-08-10 12:34 +0800
    Re: It is stunning when you see how badly Windows operates: indexing Paul <nospam@needed.invalid> - 2025-08-10 11:25 -0400
      Re: It is stunning when you see how badly Windows operates: indexing Alan <nuh-uh@nope.com> - 2025-08-10 12:37 -0700
        Re: It is stunning when you see how badly Windows operates: indexing Paul <nospam@needed.invalid> - 2025-08-10 19:46 -0400
          Re: It is stunning when you see how badly Windows operates: indexing Alan <nuh-uh@nope.com> - 2025-08-10 17:00 -0700
            Re: It is stunning when you see how badly Windows operates: indexing Paul <nospam@needed.invalid> - 2025-08-11 00:04 -0400
              Re: It is stunning when you see how badly Windows operates: indexing Marion <marion@facts.com> - 2025-08-11 06:13 +0000
                Re: It is stunning when you see how badly Windows operates: indexing Hank Rogers <invalid@nospam.com> - 2025-08-11 06:23 +0000
                Re: It is stunning when you see how badly Windows operates: indexing Alan <nuh-uh@nope.com> - 2025-08-11 11:26 -0700
                Re: It is stunning when you see how badly Windows operates: indexing Paul <nospam@needed.invalid> - 2025-08-12 01:15 -0400
                Re: It is stunning when you see how badly Windows operates: indexing Alan <nuh-uh@nope.com> - 2025-08-12 08:54 -0700
                Re: It is stunning when you see how badly Windows operates: indexing Paul <nospam@needed.invalid> - 2025-08-12 13:41 -0400
                Re: It is stunning when you see how badly Windows operates: indexing Alan <nuh-uh@nope.com> - 2025-08-12 10:54 -0700
              Re: It is stunning when you see how badly Windows operates: indexing Alan <nuh-uh@nope.com> - 2025-08-11 10:57 -0700
                Re: It is stunning when you see how badly Windows operates: indexing Alan <nuh-uh@nope.com> - 2025-08-12 08:52 -0700
      Re: It is stunning when you see how badly Windows operates: indexing "Mr. Man-wai Chang" <toylet.toylet@gmail.com> - 2025-08-11 20:22 +0800
        Re: It is stunning when you see how badly Windows operates: indexing Marion <marion@facts.com> - 2025-08-11 19:16 +0000
          Re: It is stunning when you see how badly Windows operates: indexing Alan <nuh-uh@nope.com> - 2025-08-11 17:03 -0700
          Re: It is stunning when you see how badly Windows operates: indexing "Mr. Man-wai Chang" <toylet.toylet@gmail.com> - 2025-08-12 12:36 +0800
  Re: It is stunning when you see how badly Windows operates: indexing WolfFan <akwolffan@zoho.com> - 2025-08-10 18:33 -0400
    Re: It is stunning when you see how badly Windows operates: indexing Paul <nospam@needed.invalid> - 2025-08-11 02:12 -0400
      Re: It is stunning when you see how badly Windows operates: indexing Alan <nuh-uh@nope.com> - 2025-08-11 11:25 -0700
        Re: It is stunning when you see how badly Windows operates: indexing Paul <nospam@needed.invalid> - 2025-08-11 18:02 -0400
          Re: It is stunning when you see how badly Windows operates: indexing Alan <nuh-uh@nope.com> - 2025-08-11 17:01 -0700
            Re: It is stunning when you see how badly Windows operates: indexing Paul <nospam@needed.invalid> - 2025-08-12 01:24 -0400
              Re: It is stunning when you see how badly Windows operates: indexing Alan <nuh-uh@nope.com> - 2025-08-12 08:49 -0700
              Re: It is stunning when you see how badly Windows operates: indexing Marion <marion@facts.com> - 2025-08-12 17:10 +0000
          Re: It is stunning when you see how badly Windows operates: indexing Alan <nuh-uh@nope.com> - 2025-08-12 08:55 -0700
  Re: It is stunning when you see how badly Windows operates: indexing Tom Elam <thomas.e.elam@gmail.com> - 2025-08-17 16:03 -0400
    Re: It is stunning when you see how badly Windows operates: indexing Paul <nospam@needed.invalid> - 2025-08-17 18:15 -0400

csiph-web