[Fwd: Re: best way to add unknowns from the command line?]

View: New views
2 Messages — Rating Filter:   Alert me  

[Fwd: Re: best way to add unknowns from the command line?]

by Alexander Belchenko :: Rate this Message:

| View Threaded | Show Only this Message

SOrry, forgot to hit reply-all.

-------- Исходное сообщение --------
Тема: Re: best way to add unknowns from the command line?
Дата: Sat, 31 Mar 2012 22:42:58 +0300
От: Alexander Belchenko <bialix@...>
Кому: Chris Hecker <checker@...>
Ссылки: <4F6A435E.1010706@...> <4F6AD5D0.8030103@...>
<m21uoljaf3.fsf@...> <4F6AF394.4080009@...>
<4F6AF7D9.9040309@...> <4F6D61A1.8050105@...>
<4F70266A.4020000@...>
<b73f88c7-3096-4171-b455-fd379a7b8893@...>

Chris Hecker пишет:
> I only have a few entries in unknowns, but that's because st/unknowns do
> it the scalable way, and don't recurse. One of the entries is "docs/"
> which has 2gb of binary data like images and videos, and another is
> "libs/test/" which has probabaly 10 library source trees in subdirs that
> I'm evaluating.

2gb of binary data does not matter. Only number of files matters.
You have a lot of files and subdirectories, obviously. You could
selectively ignore content of some directories, e.g.

libs/test/foo/*
libs/test/bar/*

to help qadd, but I understand that this is not solution you want,
just poor workaround.

qadd runs `bzr add` with `--no-recurse` options behind the scene to
make sure that we will add only those files user selected in GUI. So
we don't blindly add the whole directories, because we have no
guarantee that list of files does not changed between the time qadd
was invoked and OK button is pressed. Therefore qadd has to recurse
and collect the entire list of unknown files.

One can imagine that there is other smart way to do the right thing,
but so far you're the first person who complains. That does not matter
that current behavior cannot be improved, but somebody should actually
write a patch for it.

>
> Chris
>
>
> Alexander Belchenko <bialix@...> wrote:
>
>     Chris Hecker пишет:
>     > I ignore stuff I will never want to add, but I usually have a couple
>     > directories and a couple files that I keep unversioned and unignored
>     > because I'm trying to figure out what I want to do them, and I like
>     > them showing up in bzr st, which I run constantly.  It's kind of like
>     > a little todo list.
>
>     That looks weird. If you have only couple of files and couple of
>     directories then why you said qadd was very slow to open your tree with
>     couple of unknown directories?


--
All the dude wanted was his rug back


--
All the dude wanted was his rug back


Re: [Fwd: Re: best way to add unknowns from the command line?]

by Chris Hecker :: Rate this Message:

| View Threaded | Show Only this Message


Yeah, I don't use qadd so don't need a workaround for it specifically, I
was just looking at it as an option for adding unknowns easily and
selectively, but the recurse-at-startup thing was a showstopper.

A print-relative-paths parameter to status and unknowns would be best
for my use-case, but I started the thread not to necessarily ask for a
feature, just to find out what others do.

Chris


On 2012/03/31 12:43, Alexander Belchenko wrote:

> SOrry, forgot to hit reply-all.
>
> -------- Исходное сообщение --------
> Тема: Re: best way to add unknowns from the command line?
> Дата: Sat, 31 Mar 2012 22:42:58 +0300
> От: Alexander Belchenko <bialix@...>
> Кому: Chris Hecker <checker@...>
> Ссылки: <4F6A435E.1010706@...> <4F6AD5D0.8030103@...>
> <m21uoljaf3.fsf@...> <4F6AF394.4080009@...>
> <4F6AF7D9.9040309@...> <4F6D61A1.8050105@...>
> <4F70266A.4020000@...>
> <b73f88c7-3096-4171-b455-fd379a7b8893@...>
>
> Chris Hecker пишет:
>> I only have a few entries in unknowns, but that's because st/unknowns
>> do it the scalable way, and don't recurse. One of the entries is
>> "docs/" which has 2gb of binary data like images and videos, and
>> another is "libs/test/" which has probabaly 10 library source trees in
>> subdirs that I'm evaluating.
>
> 2gb of binary data does not matter. Only number of files matters.
> You have a lot of files and subdirectories, obviously. You could
> selectively ignore content of some directories, e.g.
>
> libs/test/foo/*
> libs/test/bar/*
>
> to help qadd, but I understand that this is not solution you want,
> just poor workaround.
>
> qadd runs `bzr add` with `--no-recurse` options behind the scene to
> make sure that we will add only those files user selected in GUI. So
> we don't blindly add the whole directories, because we have no
> guarantee that list of files does not changed between the time qadd
> was invoked and OK button is pressed. Therefore qadd has to recurse
> and collect the entire list of unknown files.
>
> One can imagine that there is other smart way to do the right thing,
> but so far you're the first person who complains. That does not matter
> that current behavior cannot be improved, but somebody should actually
> write a patch for it.
>
>>
>> Chris
>>
>>
>> Alexander Belchenko <bialix@...> wrote:
>>
>>     Chris Hecker пишет:
>>     > I ignore stuff I will never want to add, but I usually have a
>> couple
>>     > directories and a couple files that I keep unversioned and
>> unignored
>>     > because I'm trying to figure out what I want to do them, and I like
>>     > them showing up in bzr st, which I run constantly.  It's kind of
>> like
>>     > a little todo list.
>>
>>     That looks weird. If you have only couple of files and couple of
>>     directories then why you said qadd was very slow to open your tree
>> with     couple of unknown directories?
>
>