Re: [git-users] moving files to names that were part of the repository

[Xen]

Michael schreef op 12-08-2016 23:01:

On 2016-08-12, at 10:54 AM, Konstantin Khomoutov
 wrote:


 git config --add --global alias.rlog `log -M -C`


Ok, can you tell me what I'm doing wrong here?


You're using the wrong quotes. The backticks are going to try to execute 
a command and replace its output there on the command line in this case.


So it tries to execute the command "log" (not git log or anything) and 
it fails because log does not exist.


Regards.

(You need to use regular apostroph).

--
You received this message because you are subscribed to the Google Groups "Git for 
human beings" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to git-users+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Re: [git-users] moving files to names that were part of the repository

[Michael]

On 2016-08-12, at 10:54 AM, Konstantin Khomoutov 
 wrote:

> 1. http://www.gelato.unsw.edu.au/archives/git/0504/0598.html

Wow, that thread got big really fast...

---
Entertaining minecraft videos
http://YouTube.com/keybounce

-- 
You received this message because you are subscribed to the Google Groups "Git 
for human beings" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to git-users+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Re: [git-users] moving files to names that were part of the repository

[Michael]

On 2016-08-12, at 10:54 AM, Konstantin Khomoutov 
 wrote:

>  git config --add --global alias.rlog `log -M -C`

Ok, can you tell me what I'm doing wrong here?

keybounceMBP:Finite-Fluids michael$  git config --add --global alias.rlog `log 
-M -C`
-bash: log: command not found
error: wrong number of arguments
usage: git config []

Config file location
--global  use global config file
--system  use system config file
--local   use repository config file
-f, --file  use given config file
--blob   read config from given blob object

...

---
Entertaining minecraft videos
http://YouTube.com/keybounce

-- 
You received this message because you are subscribed to the Google Groups "Git 
for human beings" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to git-users+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Re: [git-users] moving files to names that were part of the repository

[Xen]

Konstantin Khomoutov schreef op 12-08-2016 19:54:


Depend on what you're asking for.

If you have asked about the `git log`, then yes: have an alias like

  git config --add --global alias.rlog `log -M -C`

If you have asked about whether it's possible to change the way Git
stores the history, then the answer is no.

As usually, I'd urge you to read an extensive overview of why Git
manages information in the way it does provided by its very creator 
[1].

While you certainly may disagree on the points made there and/or with
the conclusions drawn, it helps in acquiring the correct perspective: 
in

that text, some interesting areas are covered -- such as obtaining the
same information from several sources in an (octopus) merge or patches
-- things most of us never deal with.  What I'm leading to, is that Git
was created not as an academic project to create some idealized VC
system tailored for no particular real-world case, and implementing
whatever seemed "sensible" for a VC system to do.


Well, I don't care too much about other people's opinions or what they 
think should be the proper use case for a certain system, after all, my 
particular question was not at all of the academic kind, but pertained 
to a particular use case of being able to easily perceive the history of 
a file, in the sense of knowing where it came from and what has happened 
to it. I know (now) there are Git tools that will show me this 
information, likely.


I am not concerned with changing the internals of Git. If, as you say, 
Git is perfectly capable of piecing together the "real" history of a 
file without actually storing that information (directly), then from a 
practical point of view it shouldn't matter all that much, as long as it 
is usable and accessible.


I am sure the philosophical underpinnings of Git are quite proficient, 
including its fundamental "efficiency" model so to speak. But I don't 
need to know abut them today (maybe later).


First comes a use case and automated tools are fine but many of the 
higher level Git commands are not easy to use and take a while to put 
them into your system where you have all these aliases; and any system 
with a lot of aliases that are actually quite required in order for the 
system to be useful really increase the scope of the commitment you must 
have to the system in the first place.


The short way of saying that is that defaults matter, of course.

So today for me this question was really about: what do real people, 
regular people, do, when they are discontent with the level of 
information that is usually shown, and what would be their best 
practices in showing that.


I will attempt Git -M -C next time I use it, thank you.

The problem I have constantly is that I am "very bad at" saving all of 
these customizations so I have to reapply them each time I install a new 
system, so to speak, quite literally.


This just puts a bigger burden onto the usability of the system because 
those "customizations" are pretty much required for a proper use of the 
system, I feel, and this "developmental investment" is something you 
need to save.


I haven't been able to do so very well thus far, usually. They weren't 
important enough but still bite you when you don't have them :-/.



1. http://www.gelato.unsw.edu.au/archives/git/0504/0598.html


Thank you for the long text you wrote anway, relatively speaking, very 
few people take the time to write anything as thorough or complete like 
that, or anything of that kind.


So again, I do appreciate you taking the time to write that.

It's just that git is a bit of a git at times ;-). (As per the words of 
Mister Torvalds himself, I guess).


Regards :).

--
You received this message because you are subscribed to the Google Groups "Git for 
human beings" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to git-users+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Re: [git-users] moving files to names that were part of the repository

[Konstantin Khomoutov]

On Fri, 12 Aug 2016 18:59:37 +0200
Xen  wrote:

> > What `git mv` did is removed -- from the commit you have recorded
> > after running that command -- the file as it was known under its
> > old name and added the same contents under its new name.
> 
> Sometimes Git will recognise something as a rename if you do nothing 
> else in that commit, but if you merge the commit with something else, 
> suddenly it will not see or show it as that anymore.
> 
> For me personally too, renames are meaningful.
> 
> It's shame these renames are not detected or shown my by default.
> 
> Is there a best level approach to being more aware of any renames you 
> have done?
> 
> For a person seeing a delete and an add of the same file under the 
> different names is rather difficult to follow.
> 
> Is there an option to get Git to always track renames?

Depend on what you're asking for.

If you have asked about the `git log`, then yes: have an alias like

  git config --add --global alias.rlog `log -M -C`

If you have asked about whether it's possible to change the way Git
stores the history, then the answer is no.

As usually, I'd urge you to read an extensive overview of why Git
manages information in the way it does provided by its very creator [1].
While you certainly may disagree on the points made there and/or with
the conclusions drawn, it helps in acquiring the correct perspective: in
that text, some interesting areas are covered -- such as obtaining the
same information from several sources in an (octopus) merge or patches
-- things most of us never deal with.  What I'm leading to, is that Git
was created not as an academic project to create some idealized VC
system tailored for no particular real-world case, and implementing
whatever seemed "sensible" for a VC system to do.  Quite on the
contrary, Git is a very opininonated SCM tailored for the single
particular case: managing the Linux kernel.  The properties it gained
thanks to that goal (notably the blazing speed in the common case of
lots of small- to mid-sized files, and a set of sharp and powerful
tools) supposedly made it the most used SCM today, but it's still an
opinionated system, and it helps to know these "opinions" to understand
why it does things the way it does.  So take this text not as a dogma
to follow and praise but as an in-depth explanation of the design.

1. http://www.gelato.unsw.edu.au/archives/git/0504/0598.html

-- 
You received this message because you are subscribed to the Google Groups "Git 
for human beings" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to git-users+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Re: [git-users] moving files to names that were part of the repository

[Xen]

Konstantin Khomoutov schreef op 12-08-2016 18:29:


What `git mv` did is removed -- from the commit you have recorded after
running that command -- the file as it was known under its old name and
added the same contents under its new name.


Sometimes Git will recognise something as a rename if you do nothing 
else in that commit, but if you merge the commit with something else, 
suddenly it will not see or show it as that anymore.


For me personally too, renames are meaningful.

It's shame these renames are not detected or shown my by default.

Is there a best level approach to being more aware of any renames you 
have done?


For a person seeing a delete and an add of the same file under the 
different names is rather difficult to follow.


Is there an option to get Git to always track renames?

--
You received this message because you are subscribed to the Google Groups "Git for 
human beings" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to git-users+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Re: [git-users] moving files to names that were part of the repository

[Konstantin Khomoutov]

On Fri, 12 Aug 2016 17:42:20 +0200
Pablo Rodríguez  wrote:

> I’m writing a book with git as control version system.
> 
> I have been writing over 300 pages in a file named `my-book.tex`. Now
> I realize that I should start it from scratch. So I just `git mv
> my-book.tex old-draft.tex`.
> 
> The new book will have also a `my-book.tex` file. The file has been
> added to git (at I commited some changes) with the name
> `new-book.tex`.
> 
> Would it be problematic in the future if I `git mv new-book.tex
> my-book.tex`?
> 
> I assume that git history will include the name change and it won’t
> change file contents. Is that right or am I missing something?

No, the history won't specifically include any information about the
name change (Git does not explicitly track renames of the entities it
manages) but nothing bad will happen notwithstanding.
What `git mv` did is removed -- from the commit you have recorded after
running that command -- the file as it was known under its old name and
added the same contents under its new name.  So, all the old history is
still there (in the commits before the rename) and all the new history
is there as well.

The only problem this might cause is that if you will run an unadorned

  git log old-draft.tex

command, it will stop traversing the chain of commits touching
"old-draft.tex" as soon as it hits the commit in which that file was
renamed.  If you want to make `git log` still dig deeper down the
history, you need to explicitly tell it to detect renames.
Please read the `git log` manual for the --find-renames (and
--find-copies) command-line options.

In other words, Git tracks contents paying little attention to where
exactly it is located in terms of files; and if you get concerned about
locations, Git is able to detect renames and copies at the time it
traverses the recorded history of your repository.

I understand that many people have difficulty grasping this approach
but note that quite many "questions" you may ask your history are
really not concerned with files.  Say, you can ask Git to look for the
commit in which a particular word or sentence was added or removed, and
it will search through the contents, not files.
That is, Git assumes a project is a whole thing.  If you are tempted to
consider histories of individual files in the project this might be
indicative of the fact you might better have different projects for
these different files.

-- 
You received this message because you are subscribed to the Google Groups "Git 
for human beings" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to git-users+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Re: [git-users] moving files to names that were part of the repository

[Michael]

On 2016-08-12, at 8:42 AM, Pablo Rodríguez  wrote:

> Dear list,
> 
> I’m writing a book with git as control version system.
> 
> I have been writing over 300 pages in a file named `my-book.tex`. Now I
> realize that I should start it from scratch. So I just `git mv
> my-book.tex old-draft.tex`.
> 
> The new book will have also a `my-book.tex` file. The file has been
> added to git (at I commited some changes) with the name `new-book.tex`.
> 
> Would it be problematic in the future if I `git mv new-book.tex
> my-book.tex`?
> 
> I assume that git history will include the name change and it won’t
> change file contents. Is that right or am I missing something?
> 
> Many thanks for your help,
> 
> Pablo
> -- 
> http://www.ousia.tk

It sounds like you have a concern about git tracking changes in a file based on 
the name of the file.

Git actually matches files based on their content more than their name or the 
blob that they came in. While I have not seen this myself, others have told me 
that git can even tell if a file is split into two parts, and track which lines 
came from which file.

So having a file move to "old", and something else come in with the same name 
but different content will not be a problem -- git tracks the content changes 
more than the file name.


---
Entertaining minecraft videos
http://YouTube.com/keybounce

-- 
You received this message because you are subscribed to the Google Groups "Git 
for human beings" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to git-users+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.