Discussion:
commit message empty
(too old to reply)
j. van den hoff
2013-10-15 17:23:37 UTC
Permalink
I'm asking this for a colleague just starting to use fossil under windows
(with which I have no experience whatsoever):

he succeeded to clone the repo (after circumventing the previously
mentioned SSL cerificate problem) and to open it.

so he added some new file to the checkout dir and did `fossil add
new_file' and then `fossil commit'. his mail says that at this point

notpad pops up with the checkin template text and he added a message and
saved+exited the editor.

problem: he then gets the "empty commit message" warning at the command
prompt (but no error message or anything) which he confirmed and which
indeed let to a commit with an empty commit message.

I suspect this has to do with write permission to some temporary location
but don't know anything of windows etc.

any help/tip would be appreciated.

--
Using Opera's revolutionary email client: http://www.opera.com/mail/
s***@gmail.com
2013-10-15 17:39:00 UTC
Permalink
I got something similar when I inadvertently left the prefix character '#'
in front of my comment.


On Tue, Oct 15, 2013 at 1:23 PM, j. van den hoff
<***@googlemail.com>wrote:

> I'm asking this for a colleague just starting to use fossil under windows
> (with which I have no experience whatsoever):
>
> he succeeded to clone the repo (after circumventing the previously
> mentioned SSL cerificate problem) and to open it.
>
> so he added some new file to the checkout dir and did `fossil add
> new_file' and then `fossil commit'. his mail says that at this point
>
> notpad pops up with the checkin template text and he added a message and
> saved+exited the editor.
>
> problem: he then gets the "empty commit message" warning at the command
> prompt (but no error message or anything) which he confirmed and which
> indeed let to a commit with an empty commit message.
>
> I suspect this has to do with write permission to some temporary location
> but don't know anything of windows etc.
>
> any help/tip would be appreciated.
>
> --
> Using Opera's revolutionary email client: http://www.opera.com/mail/
> ______________________________**_________________
> fossil-users mailing list
> fossil-***@lists.fossil-scm.**org <fossil-***@lists.fossil-scm.org>
> http://lists.fossil-scm.org:**8080/cgi-bin/mailman/listinfo/**fossil-users<http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users>
>
j. van den hoff
2013-10-15 19:05:52 UTC
Permalink
On Tue, 15 Oct 2013 19:39:00 +0200, <***@gmail.com> wrote:

> I got something similar when I inadvertently left the prefix character
> '#'
> in front of my comment.

thanks a lot! would never have thought of that but apparently (if I've
understood his answer correctly) he did just that on purpose, presuming
that that's the way the commit message should look like (sounds like a
good candidate for the FAQ ...).

>
>
> On Tue, Oct 15, 2013 at 1:23 PM, j. van den hoff
> <***@googlemail.com>wrote:
>
>> I'm asking this for a colleague just starting to use fossil under
>> windows
>> (with which I have no experience whatsoever):
>>
>> he succeeded to clone the repo (after circumventing the previously
>> mentioned SSL cerificate problem) and to open it.
>>
>> so he added some new file to the checkout dir and did `fossil add
>> new_file' and then `fossil commit'. his mail says that at this point
>>
>> notpad pops up with the checkin template text and he added a message and
>> saved+exited the editor.
>>
>> problem: he then gets the "empty commit message" warning at the command
>> prompt (but no error message or anything) which he confirmed and which
>> indeed let to a commit with an empty commit message.
>>
>> I suspect this has to do with write permission to some temporary
>> location
>> but don't know anything of windows etc.
>>
>> any help/tip would be appreciated.
>>
>> --
>> Using Opera's revolutionary email client: http://www.opera.com/mail/
>> ______________________________**_________________
>> fossil-users mailing list
>> fossil-***@lists.fossil-scm.**org <fossil-***@lists.fossil-scm.org>
>> http://lists.fossil-scm.org:**8080/cgi-bin/mailman/listinfo/**fossil-users<http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users>
>>


--
Using Opera's revolutionary email client: http://www.opera.com/mail/
s***@gmail.com
2013-10-15 20:58:16 UTC
Permalink
Haha, yeah that is one of the many in my fossil cheat sheet. ;)
I'm still on the hook to deliver that. Just need to sanitize and prettify.
Right now it is combined with my Windows Batch/Doskey commands.
I was hoping to replace all that with the fossil api.


On Tue, Oct 15, 2013 at 3:05 PM, j. van den hoff
<***@googlemail.com>wrote:

> On Tue, 15 Oct 2013 19:39:00 +0200, <***@gmail.com> wrote:
>
> I got something similar when I inadvertently left the prefix character '#'
>> in front of my comment.
>>
>
> thanks a lot! would never have thought of that but apparently (if I've
> understood his answer correctly) he did just that on purpose, presuming
> that that's the way the commit message should look like (sounds like a good
> candidate for the FAQ ...).
>
>
>>
>> On Tue, Oct 15, 2013 at 1:23 PM, j. van den hoff
>> <***@googlemail.com>**wrote:
>>
>> I'm asking this for a colleague just starting to use fossil under windows
>>> (with which I have no experience whatsoever):
>>>
>>> he succeeded to clone the repo (after circumventing the previously
>>> mentioned SSL cerificate problem) and to open it.
>>>
>>> so he added some new file to the checkout dir and did `fossil add
>>> new_file' and then `fossil commit'. his mail says that at this point
>>>
>>> notpad pops up with the checkin template text and he added a message and
>>> saved+exited the editor.
>>>
>>> problem: he then gets the "empty commit message" warning at the command
>>> prompt (but no error message or anything) which he confirmed and which
>>> indeed let to a commit with an empty commit message.
>>>
>>> I suspect this has to do with write permission to some temporary location
>>> but don't know anything of windows etc.
>>>
>>> any help/tip would be appreciated.
>>>
>>> --
>>> Using Opera's revolutionary email client: http://www.opera.com/mail/
>>> ______________________________****_________________
>>> fossil-users mailing list
>>> fossil-***@lists.fossil-scm.****org <fossil-***@lists.fossil-**
>>> scm.org <fossil-***@lists.fossil-scm.org>>
>>> http://lists.fossil-scm.org:****8080/cgi-bin/mailman/listinfo/**
>>> **fossil-users<http://lists.**fossil-scm.org:8080/cgi-bin/**
>>> mailman/listinfo/fossil-users<http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users>
>>> >
>>>
>>>
>
> --
> Using Opera's revolutionary email client: http://www.opera.com/mail/
> ______________________________**_________________
> fossil-users mailing list
> fossil-***@lists.fossil-scm.**org <fossil-***@lists.fossil-scm.org>
> http://lists.fossil-scm.org:**8080/cgi-bin/mailman/listinfo/**fossil-users<http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users>
>
Stephan Beal
2013-10-16 11:16:39 UTC
Permalink
On Tue, Oct 15, 2013 at 9:05 PM, j. van den hoff
<***@googlemail.com>wrote:

> the way the commit message should look like (sounds like a good candidate
> for the FAQ ...).
>

(to be pedantic for a moment (surprise! ;))... except that it fails the "F"
part of "FAQ" ;).

But yeah, this should be mentioned somewhere. Suggestions are welcomed. How
about simply to the basic commit help?

[***@host:~]$ f help commit
Usage: f commit ?OPTIONS? ?FILE...?

Create a new version containing all of the changes in the current
checkout. You will be prompted to enter a check-in comment unless
the comment has been specified on the command-line using "-m" or a
file containing the comment using -M. The editor defined in the
"editor" fossil option (see f help set) will be used, or from
the "VISUAL" or "EDITOR" environment variables (in that order) if
no editor is set.
<<<here somewhere>>>

?

--
----- stephan beal
http://wanderinghorse.net/home/stephan/
http://gplus.to/sgbeal
"Since tyranny's the only guaranteed byproduct of those who insist on a
perfect world, freedom will have to do." -- Bigby Wolf
j. van den hoff
2013-10-16 11:31:26 UTC
Permalink
On Wed, 16 Oct 2013 13:16:39 +0200, Stephan Beal <***@googlemail.com>
wrote:

> On Tue, Oct 15, 2013 at 9:05 PM, j. van den hoff
> <***@googlemail.com>wrote:
>
>> the way the commit message should look like (sounds like a good
>> candidate
>> for the FAQ ...).
>>
>
> (to be pedantic for a moment (surprise! ;))... except that it fails the
> "F"
> part of "FAQ" ;).
>
> But yeah, this should be mentioned somewhere. Suggestions are welcomed.
> How
> about simply to the basic commit help?
>
> [***@host:~]$ f help commit
> Usage: f commit ?OPTIONS? ?FILE...?
>
> Create a new version containing all of the changes in the current
> checkout. You will be prompted to enter a check-in comment unless
> the comment has been specified on the command-line using "-m" or a
> file containing the comment using -M. The editor defined in the
> "editor" fossil option (see f help set) will be used, or from
> the "VISUAL" or "EDITOR" environment variables (in that order) if
> no editor is set.
> <<<here somewhere>>>

yes, that would be sensible, I believe, although
actually, I've just "discovered" that it _is_ already stated in the commit
template itself:

"# Enter comments on this check-in. Lines beginning with # are ignored."

so this definitely _should_ have sufficed to notify the unwary user but it
did not (and then the help text probably is ignored, too...): in my case
the colleague in question seemingly took it for granted that the word
"comment" in the template text meant "shell-type comment (i.e. starting
with hash)) so he quite intentionally started his "comment" in this way.

so maybe that text better should be rephrased to something like:

"# Enter your commit message for this check-in above this line. Lines
beginning with # are ignored."


another question/idea: would it not be relatively easy to change the
behaviour to:

"# Enter your commit message for this check-in above this line. This line
and everything below is ignored."

meaning that this specific pattern (the whole, exact line) determines end
of the commit message (which then very well might include #-lines if the
user so desires)?
>
> ?
>


--
Using Opera's revolutionary email client: http://www.opera.com/mail/
Stephan Beal
2013-10-16 12:06:35 UTC
Permalink
On Wed, Oct 16, 2013 at 1:31 PM, j. van den hoff
<***@googlemail.com>wrote:

> actually, I've just "discovered" that it _is_ already stated in the commit
> template itself:
>
> "# Enter comments on this check-in. Lines beginning with # are ignored."
>
> so this definitely _should_ have sufficed to notify the unwary user but it
> did not (and then the help text probably is ignored, too...):


Right -anyone who doesn't read that text is not going to look at the help.
There's not much we can do about that, and the result of ignoring it is
harmless - an empty commit message which can then be edited after-the-fact.

so maybe that text better should be rephrased to something like:
>
> "# Enter your commit message for this check-in above this line. Lines
> beginning with # are ignored."
>

Agreed - i'll change "comment" to "message" (unless a better suggestion
comes up before i get it changed). i'm still at work, but will get this
done in the next couple days. Tomorrow and Friday i'm working from home, so
i'll have some extra time to play around (@boss: by that i mean only "no
travel time" :).


> another question/idea: would it not be relatively easy to change the
> behaviour to:
>
> "# Enter your commit message for this check-in above this line. This line
> and everything below is ignored."


> meaning that this specific pattern (the whole, exact line) determines end
> of the commit message (which then very well might include #-lines if the
> user so desires)?


i think that would just leave a bug-in-waiting which we'd eventually trip
over (and end up with more comment text than desired, as opposed to an
empty one). e.g. when libfossil eventually gets this behaviour, someone
will want to translate that text, and then an exact match will break. The
current convention has a long and glorious history in svn, cvs, etc., and
is easier to implement (is already implemented!) so i'd rather just stick
with that.

--
----- stephan beal
http://wanderinghorse.net/home/stephan/
http://gplus.to/sgbeal
"Since tyranny's the only guaranteed byproduct of those who insist on a
perfect world, freedom will have to do." -- Bigby Wolf
Stephan Beal
2013-10-16 12:50:10 UTC
Permalink
On Wed, Oct 16, 2013 at 2:06 PM, Stephan Beal <***@googlemail.com> wrote:

> Agreed - i'll change "comment" to "message" (unless a better suggestion
> comes up before i get it changed). i'm still at work, but will get this
> done in the next couple days. Tomorrow and Friday i'm working from home, so
> i'll have some extra time to play around (@boss: by that i mean only "no
> travel time" :).
>

Okay, i get to wait on a 770MB file to copy to China at ~160kb per second,
so...

This is what i see in my $EDITOR (emacs) right now...


Changed text which appears in EDITOR/VISUAL-based commit message editor,
per ML discussion.
# Enter commit message for this check-in. Lines beginning with # are
ignored.
#
# user: stephan
# tags: trunk
#
# EDITED src/checkin.c


i didn't touch the basic help text because, after trying it, it just seemed
out of place and like "too much info," considering that 'help commit'
already has so much text and this addition would so rarely be
informative/useful at that level. Right now it's shown at the one exact
place where it's relevant, which seems most useful to me.

So...

ctrl-x ctrl-s ctrl-x c
(no lip from the vi users, please ;)
...

http://fossil-scm.org/index.html/info/9801449b68

We could probably shorten "commit message" to "message" or change it to
"change summary" - suggestions are welcomed, but it ideally should fit in
(approx.) 72 characters (only for historical reasons).

--
----- stephan beal
http://wanderinghorse.net/home/stephan/
http://gplus.to/sgbeal
"Since tyranny's the only guaranteed byproduct of those who insist on a
perfect world, freedom will have to do." -- Bigby Wolf
s***@gmail.com
2013-10-16 14:33:41 UTC
Permalink
When I stumbled on this '#' comment character commit issue, I was inclined
to ask why that was chosen? I actually lost a '#define blah blah' comment I
wanted to highlight.
Can the comment character be user selectable at this late stage?
I would prefer comments lead with ! or ' or ; or // or /* or ;=.

#Thanks for fossil!#
Stephan Beal
2013-10-16 14:48:37 UTC
Permalink
On Wed, Oct 16, 2013 at 4:33 PM, <***@gmail.com> wrote:

> When I stumbled on this '#' comment character commit issue, I was inclined
> to ask why that was chosen? I actually lost a '#define blah blah' comment I
> wanted to highlight.
> Can the comment character be user selectable at this late stage?
> I would prefer comments lead with ! or ' or ; or // or /* or ;=.
>

Considering the relatively harmless effect of losing an unusual (though
admittedly valid) commit message (which can be edited afterwards), and the
long-standing conventions surrounding the current convention (cvs, svn,
git, etc.), i won't volunteer to make that change ;). But maybe someone
else will.

Making something (anything) configurable always adds a relatively high
amount of overhead compared to hard-coding it, and this particular case
would bring a great deal more (relatively speaking) than many cases because
we'd be replacing a single hard-coded '#' with the logic to store, load,
validate (that it's exactly 1 ASCII byte), and compare the character (and
then hope someone doesn't set it to a newline or some such). As soon as
it's configurable, someone would want some Cyrillic or Arabic character as
the marker (and then they'd want RTL[1] support for the Arabic version)...
it all leads down dark, scary, paths riddled with corner cases just for the
sake of a harmless one-time loss of a commit message (which can then be
edited to fix it).

There are other use cases where '#' is useful, e.g. Google Code uses #XXXX
to refer to/detect ticket numbers, but fossil has always used [uuid] for
that.

[1] right-to-left, http://en.wikipedia.org/wiki/Right-to-left

--
----- stephan beal
http://wanderinghorse.net/home/stephan/
http://gplus.to/sgbeal
"Since tyranny's the only guaranteed byproduct of those who insist on a
perfect world, freedom will have to do." -- Bigby Wolf
j. van den hoff
2013-10-16 14:50:24 UTC
Permalink
On Wed, 16 Oct 2013 14:50:10 +0200, Stephan Beal <***@googlemail.com>
wrote:

> On Wed, Oct 16, 2013 at 2:06 PM, Stephan Beal <***@googlemail.com>
> wrote:
>
>> Agreed - i'll change "comment" to "message" (unless a better suggestion
>> comes up before i get it changed). i'm still at work, but will get this
>> done in the next couple days. Tomorrow and Friday i'm working from
>> home, so
>> i'll have some extra time to play around (@boss: by that i mean only "no
>> travel time" :).
>>
>
> Okay, i get to wait on a 770MB file to copy to China at ~160kb per
> second,
> so...
>
> This is what i see in my $EDITOR (emacs) right now...
>
>
> Changed text which appears in EDITOR/VISUAL-based commit message editor,
> per ML discussion.
> # Enter commit message for this check-in. Lines beginning with # are
> ignored.
> #
> # user: stephan
> # tags: trunk
> #
> # EDITED src/checkin.c
>
>
> i didn't touch the basic help text because, after trying it, it just
> seemed
> out of place and like "too much info," considering that 'help commit'
> already has so much text and this addition would so rarely be
> informative/useful at that level. Right now it's shown at the one exact
> place where it's relevant, which seems most useful to me.

+1

>
> So...
>
> ctrl-x ctrl-s ctrl-x c
> (no lip from the vi users, please ;)

I just was going to build up some steam ;-). anyway, I believe that's
where emcacs's gotten it's name from: "EscapeMetaAlternateControlShift" (3
of which, I presume, are the minimum number of keys you have to press at
the same time -- plus some letter ;-))
> ...
>
> http://fossil-scm.org/index.html/info/9801449b68
>
> We could probably shorten "commit message" to "message" or change it to

I think the `commit' should be kept for clarity.

> "change summary" - suggestions are welcomed, but it ideally should fit in
> (approx.) 72 characters (only for historical reasons).

well, even the punchards had 80 characters also only
>


--
Using Opera's revolutionary email client: http://www.opera.com/mail/
Stephan Beal
2013-10-16 14:58:54 UTC
Permalink
On Wed, Oct 16, 2013 at 4:50 PM, j. van den hoff
<***@googlemail.com>wrote:

> On Wed, 16 Oct 2013 14:50:10 +0200, Stephan Beal <***@googlemail.com>
> wrote:
>
> On Wed, Oct 16, 2013 at 2:06 PM, Stephan Beal <***@googlemail.com>
>> wrote:
>>
> ctrl-x ctrl-s ctrl-x c
>> (no lip from the vi users, please ;)
>>
>
> I just was going to build up some steam ;-). anyway, I believe that's
> where emcacs's gotten it's name from: "**EscapeMetaAlternateControlShif**t"
> (3 of which, I presume, are the minimum number of keys you have to press at
> the same time -- plus some letter ;-))


i've found that you can often determine whether someone is an emacs user by
looking at their keyboard. On all of mine the left ctrl, 'X', and 'S' keys
are used so often that the letters have been worn away.

...
>
>>
>> http://fossil-scm.org/index.**html/info/9801449b68<http://fossil-scm.org/index.html/info/9801449b68>
>>
>> We could probably shorten "commit message" to "message" or change it to
>>
>
> I think the `commit' should be kept for clarity.


Agreed.


>
> "change summary" - suggestions are welcomed, but it ideally should fit in
>> (approx.) 72 characters (only for historical reasons).
>>
>
> well, even the punchards had 80 characters also only


True, i'm just thinking about word wrap within the $EDITOR itself. We don't
want that some auto-wrap feature wraps that line by accident, which could
truly mess up the commit message. (PS: i'm not _quite_ old enough to have
used punchcards. i've seen them but never used them.)

--
----- stephan beal
http://wanderinghorse.net/home/stephan/
http://gplus.to/sgbeal
"Since tyranny's the only guaranteed byproduct of those who insist on a
perfect world, freedom will have to do." -- Bigby Wolf
s***@gmail.com
2013-10-16 16:57:11 UTC
Permalink
Yeah, I only asked for a user defined comment character since I thought it
too bold to request it be changed outright. '#' is really too useful to me
to be relegated to comments :(
So many other useless characters to choose from?


On Wed, Oct 16, 2013 at 10:58 AM, Stephan Beal <***@googlemail.com>wrote:

> On Wed, Oct 16, 2013 at 4:50 PM, j. van den hoff <
> ***@googlemail.com> wrote:
>
>> On Wed, 16 Oct 2013 14:50:10 +0200, Stephan Beal <***@googlemail.com>
>> wrote:
>>
>> On Wed, Oct 16, 2013 at 2:06 PM, Stephan Beal <***@googlemail.com>
>>> wrote:
>>>
>> ctrl-x ctrl-s ctrl-x c
>>> (no lip from the vi users, please ;)
>>>
>>
>> I just was going to build up some steam ;-). anyway, I believe that's
>> where emcacs's gotten it's name from: "**EscapeMetaAlternateControlShif**t"
>> (3 of which, I presume, are the minimum number of keys you have to press at
>> the same time -- plus some letter ;-))
>>
>
> i've found that you can often determine whether someone is an emacs user
> by looking at their keyboard. On all of mine the left ctrl, 'X', and 'S'
> keys are used so often that the letters have been worn away.
>
> ...
>>
>>>
>>> http://fossil-scm.org/index.**html/info/9801449b68<http://fossil-scm.org/index.html/info/9801449b68>
>>>
>>> We could probably shorten "commit message" to "message" or change it to
>>>
>>
>> I think the `commit' should be kept for clarity.
>
>
> Agreed.
>
>
>>
>> "change summary" - suggestions are welcomed, but it ideally should fit in
>>> (approx.) 72 characters (only for historical reasons).
>>>
>>
>> well, even the punchards had 80 characters also only
>
>
> True, i'm just thinking about word wrap within the $EDITOR itself. We
> don't want that some auto-wrap feature wraps that line by accident, which
> could truly mess up the commit message. (PS: i'm not _quite_ old enough to
> have used punchcards. i've seen them but never used them.)
>
> --
> ----- stephan beal
> http://wanderinghorse.net/home/stephan/
> http://gplus.to/sgbeal
> "Since tyranny's the only guaranteed byproduct of those who insist on a
> perfect world, freedom will have to do." -- Bigby Wolf
>
> _______________________________________________
> fossil-users mailing list
> fossil-***@lists.fossil-scm.org
> http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
>
>
Stephan Beal
2013-10-16 16:59:54 UTC
Permalink
On Wed, Oct 16, 2013 at 6:57 PM, <***@gmail.com> wrote:

> Yeah, I only asked for a user defined comment character since I thought it
> too bold to request it be changed outright. '#' is really too useful to me
> to be relegated to comments :(
> So many other useless characters to choose from?
>

If you follow the sqlite list you might remember that someone recently
complained that an update to the ICU library/data files breaks sqlite's
understanding of the "Mongolian Vowel Separator" character (whatever that
is). Certainly someone out there would try to use that as a common marker!
i suspect we're far better off just keeping '#' hard-coded!

--
----- stephan beal
http://wanderinghorse.net/home/stephan/
http://gplus.to/sgbeal
"Since tyranny's the only guaranteed byproduct of those who insist on a
perfect world, freedom will have to do." -- Bigby Wolf
Matt Welland
2013-10-16 18:43:27 UTC
Permalink
Simply make the ignore prefix something extremely unlikely to collide such
as:

##-fossil-## Enter comments on this check-in.
##-fossil-## Lines beginning with "##-fossil-##" are ignored.
##-fossil-##
##-fossil-## user: matt
##-fossil-## tags: v1.55
##-fossil-##
##-fossil-## EDITED db.scm
##-fossil-## EDITED tests/Makefile



On Wed, Oct 16, 2013 at 9:59 AM, Stephan Beal <***@googlemail.com> wrote:

> On Wed, Oct 16, 2013 at 6:57 PM, <***@gmail.com> wrote:
>
>> Yeah, I only asked for a user defined comment character since I thought
>> it too bold to request it be changed outright. '#' is really too useful to
>> me to be relegated to comments :(
>> So many other useless characters to choose from?
>>
>
> If you follow the sqlite list you might remember that someone recently
> complained that an update to the ICU library/data files breaks sqlite's
> understanding of the "Mongolian Vowel Separator" character (whatever that
> is). Certainly someone out there would try to use that as a common marker!
> i suspect we're far better off just keeping '#' hard-coded!
>
> --
> ----- stephan beal
> http://wanderinghorse.net/home/stephan/
> http://gplus.to/sgbeal
> "Since tyranny's the only guaranteed byproduct of those who insist on a
> perfect world, freedom will have to do." -- Bigby Wolf
>
> _______________________________________________
> fossil-users mailing list
> fossil-***@lists.fossil-scm.org
> http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
>
>


--
Matt
-=-
90% of the nations wealth is held by 2% of the people. Bummer to be in the
majority...
Stephan Beal
2013-10-16 18:44:38 UTC
Permalink
On Wed, Oct 16, 2013 at 8:43 PM, Matt Welland <***@gmail.com> wrote:

> Simply make the ignore prefix something extremely unlikely to collide such
> as:
>

Until today, '#' fell into the category of "extremely unlikely" to cause a
problem ;)

--
----- stephan beal
http://wanderinghorse.net/home/stephan/
http://gplus.to/sgbeal
"Since tyranny's the only guaranteed byproduct of those who insist on a
perfect world, freedom will have to do." -- Bigby Wolf
Ron Wilson
2013-10-16 18:51:48 UTC
Permalink
On Wed, Oct 16, 2013 at 2:43 PM, Matt Welland <***@gmail.com> wrote:

> Simply make the ignore prefix something extremely unlikely to collide such
> as:
>
> ##-fossil-## Enter comments on this check-in.
> ##-fossil-## Lines beginning with "##-fossil-##" are ignored.
> ##-fossil-##
> ##-fossil-## user: matt
> ##-fossil-## tags: v1.55
> ##-fossil-##
> ##-fossil-## EDITED db.scm
> ##-fossil-## EDITED tests/Makefile
>

I think that's too long, especially when you un-comment lines to include
them (unless your editor supports rectangular selection).
Matt Welland
2013-10-17 00:30:30 UTC
Permalink
1. "Until today, '#' fell into the category of "extremely unlikely" to
cause a problem ;)"
=> I've hit this limitation a couple times. It seemed a little silly to
me to use
a single hash in this context but since I could go into the UI and
"fix" things
I didn't consider it important enough to report.

2. "I think that's too long, especially when you un-comment lines
to include them (unless your editor supports rectangular selection)."
=> Make it shorter, how about "#fsl#"? Any decent modern editor has
rectangular
selection I'm sure. Us antiquated vi and (x)emacs users are covered of
course :)

When I have a multi-line comment to enter I just put my -m at the end of
line and use quotes however I generally prefer succinct one line comments.



On Wed, Oct 16, 2013 at 11:51 AM, Ron Wilson <***@gmail.com> wrote:

> On Wed, Oct 16, 2013 at 2:43 PM, Matt Welland <***@gmail.com> wrote:
>
>> Simply make the ignore prefix something extremely unlikely to collide
>> such as:
>>
>> ##-fossil-## Enter comments on this check-in.
>> ##-fossil-## Lines beginning with "##-fossil-##" are ignored.
>> ##-fossil-##
>> ##-fossil-## user: matt
>> ##-fossil-## tags: v1.55
>> ##-fossil-##
>> ##-fossil-## EDITED db.scm
>> ##-fossil-## EDITED tests/Makefile
>>
>
> I think that's too long, especially when you un-comment lines to include
> them (unless your editor supports rectangular selection).
>
>
> _______________________________________________
> fossil-users mailing list
> fossil-***@lists.fossil-scm.org
> http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
>
>


--
Matt
-=-
90% of the nations wealth is held by 2% of the people. Bummer to be in the
majority...
Stestagg
2013-10-17 13:14:10 UTC
Permalink
Using a marker line would allow for commit messages to contain any content

For example:
==========
[commit message goes here, any character accepted]

### End comment

Enter comments on this check-in. Lines below the last '### End comment'
are ignored

user: <user>
tags: trunk

EDITED ...
============

Thanks

Steve


On Thu, Oct 17, 2013 at 1:30 AM, Matt Welland <***@gmail.com> wrote:

> 1. "Until today, '#' fell into the category of "extremely unlikely" to
> cause a problem ;)"
> => I've hit this limitation a couple times. It seemed a little silly to
> me to use
> a single hash in this context but since I could go into the UI and
> "fix" things
> I didn't consider it important enough to report.
>
> 2. "I think that's too long, especially when you un-comment lines
>
> to include them (unless your editor supports rectangular selection)."
> => Make it shorter, how about "#fsl#"? Any decent modern editor has
> rectangular
> selection I'm sure. Us antiquated vi and (x)emacs users are covered
> of course :)
>
> When I have a multi-line comment to enter I just put my -m at the end of
> line and use quotes however I generally prefer succinct one line comments.
>
>
>
> On Wed, Oct 16, 2013 at 11:51 AM, Ron Wilson <***@gmail.com> wrote:
>
>> On Wed, Oct 16, 2013 at 2:43 PM, Matt Welland <***@gmail.com>wrote:
>>
>>> Simply make the ignore prefix something extremely unlikely to collide
>>> such as:
>>>
>>> ##-fossil-## Enter comments on this check-in.
>>> ##-fossil-## Lines beginning with "##-fossil-##" are ignored.
>>> ##-fossil-##
>>> ##-fossil-## user: matt
>>> ##-fossil-## tags: v1.55
>>> ##-fossil-##
>>> ##-fossil-## EDITED db.scm
>>> ##-fossil-## EDITED tests/Makefile
>>>
>>
>> I think that's too long, especially when you un-comment lines to include
>> them (unless your editor supports rectangular selection).
>>
>>
>> _______________________________________________
>> fossil-users mailing list
>> fossil-***@lists.fossil-scm.org
>> http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
>>
>>
>
>
> --
> Matt
> -=-
> 90% of the nations wealth is held by 2% of the people. Bummer to be in the
> majority...
>
> _______________________________________________
> fossil-users mailing list
> fossil-***@lists.fossil-scm.org
> http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
>
>
Stephan Beal
2013-10-17 13:31:37 UTC
Permalink
On Thu, Oct 17, 2013 at 3:14 PM, Stestagg <***@gmail.com> wrote:

> Using a marker line would allow for commit messages to contain any content
>

i can't say that i would like commit messages to be able to contain "any
content." They're not intended to be general purposes documentation, but
_brief_ descriptions of changes made in the context of the current commit.
A commit message field is not intended to hold a whole changelog for a
release nor a combined list of all changes from commits imported en masse
from another SCM (for such uses, just keep the change list as a separate
file and check it in along with the commit).

IMO the whole thing is far more trouble than it's worth and introduces many
corner cases and bugs where we currently don't have any (or only have very
old, well-understood ones which we can blame on the other systems which
were emulated ;). "If it ain't broke, don't touch it."

--
----- stephan beal
http://wanderinghorse.net/home/stephan/
http://gplus.to/sgbeal
"Since tyranny's the only guaranteed byproduct of those who insist on a
perfect world, freedom will have to do." -- Bigby Wolf
j. van den hoff
2013-10-17 13:55:50 UTC
Permalink
my final 2c:

before my colleague tripped over this by _assuming_ that "comment" in the
checkin template meant "shell (or perl, python, whatever) type comment"
and, due to this misconception _intentionally_ formatted his commit
message with leading `#' I was not aware of the potential problem (never
happpened to me, anyway). another poster mentioned to trip over the same
problem by referring to "#include" (or similar) in his commit message,
which I presume _might_ happen by and then to the unwary user, too.

still I agree that the sanest way is to keep it simple and identify
ignored lines by some leading short ring. here's what mercurial does:

8<----------------------------------------------------


HG: Enter commit message. Lines beginning with 'HG:' are removed.
HG: Leave message empty to abort commit.
HG: --
HG: user: someone
HG: branch 'default'
HG: added tt
8<----------------------------------------------------

I'd say that is more reaonsable than `#' (and to abort the checkin when
the message is left empty is, too, but that's another story...). so what
about -- for one -- mimicking `mercurial' here and to use

`FOSSIL:' or even `FOS:' as the unique identifier (which mostly is what
somebody else already proposed). I'd say that will work better than `#'
(i.e. accidentally names clash very very unlikely) and sufficiently stands
out on the respective lines due to the capitalization.



On Thu, 17 Oct 2013 15:31:37 +0200, Stephan Beal <***@googlemail.com>
wrote:

> On Thu, Oct 17, 2013 at 3:14 PM, Stestagg <***@gmail.com> wrote:
>
>> Using a marker line would allow for commit messages to contain any
>> content
>>
>
> i can't say that i would like commit messages to be able to contain "any
> content." They're not intended to be general purposes documentation, but
> _brief_ descriptions of changes made in the context of the current
> commit.
> A commit message field is not intended to hold a whole changelog for a
> release nor a combined list of all changes from commits imported en masse
> from another SCM (for such uses, just keep the change list as a separate
> file and check it in along with the commit).
>
> IMO the whole thing is far more trouble than it's worth and introduces
> many
> corner cases and bugs where we currently don't have any (or only have
> very
> old, well-understood ones which we can blame on the other systems which
> were emulated ;). "If it ain't broke, don't touch it."
>


--
Using Opera's revolutionary email client: http://www.opera.com/mail/
Stephan Beal
2013-10-17 13:59:45 UTC
Permalink
On Thu, Oct 17, 2013 at 3:55 PM, j. van den hoff
<***@googlemail.com>wrote:

> `FOSSIL:' or even `FOS:' as the unique identifier (which mostly is what
> somebody else already proposed). I'd say that will work better than `#'
> (i.e. accidentally names clash very very unlikely) and sufficiently stands
> out on the respective lines due to the capitalization.


Historical note (i don't think you've been on the list long enough to have
seen this): here's the summary from some code comments in fossil:

/*
** Locate the root directory of the local repository tree. The root
** directory is found by searching for a file named "_FOSSIL_" or
".fslckout"
** that contains a valid repository database.
**
** For legacy, also look for ".fos". The use of ".fos" is deprecated
** since "fos" has negative connotations in Hungarian, we are told.
...

A user made us aware at some point that fos is an ugly/inappropriate word
in his language (i concur with the code comment that it was Hungarian).

--
----- stephan beal
http://wanderinghorse.net/home/stephan/
http://gplus.to/sgbeal
"Since tyranny's the only guaranteed byproduct of those who insist on a
perfect world, freedom will have to do." -- Bigby Wolf
j. van den hoff
2013-10-17 14:03:50 UTC
Permalink
On Thu, 17 Oct 2013 15:59:45 +0200, Stephan Beal <***@googlemail.com>
wrote:

> On Thu, Oct 17, 2013 at 3:55 PM, j. van den hoff
> <***@googlemail.com>wrote:
>
>> `FOSSIL:' or even `FOS:' as the unique identifier (which mostly is what
>> somebody else already proposed). I'd say that will work better than `#'
>> (i.e. accidentally names clash very very unlikely) and sufficiently
>> stands
>> out on the respective lines due to the capitalization.
>
>
> Historical note (i don't think you've been on the list long enough to
> have
> seen this): here's the summary from some code comments in fossil:
>
> /*
> ** Locate the root directory of the local repository tree. The root
> ** directory is found by searching for a file named "_FOSSIL_" or
> ".fslckout"
> ** that contains a valid repository database.
> **
> ** For legacy, also look for ".fos". The use of ".fos" is deprecated
> ** since "fos" has negative connotations in Hungarian, we are told.
> ...
>
> A user made us aware at some point that fos is an ugly/inappropriate word
> in his language (i concur with the code comment that it was Hungarian).

well, there are _towns_ having offending names

http://en.wikipedia.org/wiki/Fucking,_Austria :-)

so I presume the hungarian users will probably don't take real offense but
anyway "FOSSIL:" would be fine to.

>


--
Using Opera's revolutionary email client: http://www.opera.com/mail/
Stephan Beal
2013-10-17 14:22:30 UTC
Permalink
On Thu, Oct 17, 2013 at 4:03 PM, j. van den hoff
<***@googlemail.com>wrote:

> well, there are _towns_ having offending names
>

lol! There's a bus line (from Germany) which has got a nearly identical
name.

"FOSSIL:" seems reasonable to me, but i'm not going to volunteer to change
it for the simple reason that have only used the $EDITOR feature one time
ever (across all SCMs, going back about 16 years), and that was to test the
"comment"/"message" change you submitted a few days ago.

For those who don't know about it: fossil supports a -M (big emm) option
which reads the commit message from a file and doesn't not apply any
#-related special handling to the content (or it didn't at the time it was
implemented, and i'm assuming that hasn't changed), e.g.:

fossil commit -M messagefile file1 file2...

--
----- stephan beal
http://wanderinghorse.net/home/stephan/
http://gplus.to/sgbeal
"Since tyranny's the only guaranteed byproduct of those who insist on a
perfect world, freedom will have to do." -- Bigby Wolf
s***@gmail.com
2013-10-17 14:23:37 UTC
Permalink
I also take exception to this being only a recent phenomena. I have been
burned several times now, but as you say, this is not a mission critical
error.
However, me and my users cherish the timeline view and are confused by
random omissions of '#'this or '#'that. So I do not feel the comments are
trivial.

I would really appreciate any of the other suggestions.
Multi-char lead being less onerous.

#Thanks for fossil!


On Thu, Oct 17, 2013 at 10:03 AM, j. van den hoff <***@googlemail.com
> wrote:

> On Thu, 17 Oct 2013 15:59:45 +0200, Stephan Beal <***@googlemail.com>
> wrote:
>
> On Thu, Oct 17, 2013 at 3:55 PM, j. van den hoff
>> <***@googlemail.com>**wrote:
>>
>> `FOSSIL:' or even `FOS:' as the unique identifier (which mostly is what
>>> somebody else already proposed). I'd say that will work better than `#'
>>> (i.e. accidentally names clash very very unlikely) and sufficiently
>>> stands
>>> out on the respective lines due to the capitalization.
>>>
>>
>>
>> Historical note (i don't think you've been on the list long enough to have
>> seen this): here's the summary from some code comments in fossil:
>>
>> /*
>> ** Locate the root directory of the local repository tree. The root
>> ** directory is found by searching for a file named "_FOSSIL_" or
>> ".fslckout"
>> ** that contains a valid repository database.
>> **
>> ** For legacy, also look for ".fos". The use of ".fos" is deprecated
>> ** since "fos" has negative connotations in Hungarian, we are told.
>> ...
>>
>> A user made us aware at some point that fos is an ugly/inappropriate word
>> in his language (i concur with the code comment that it was Hungarian).
>>
>
> well, there are _towns_ having offending names
>
> http://en.wikipedia.org/wiki/**Fucking,_Austria<http://en.wikipedia.org/wiki/Fucking,_Austria>:-)
>
> so I presume the hungarian users will probably don't take real offense but
> anyway "FOSSIL:" would be fine to.
>
>
>
>>
>
> --
> Using Opera's revolutionary email client: http://www.opera.com/mail/
> ______________________________**_________________
> fossil-users mailing list
> fossil-***@lists.fossil-scm.**org <fossil-***@lists.fossil-scm.org>
> http://lists.fossil-scm.org:**8080/cgi-bin/mailman/listinfo/**fossil-users<http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users>
>
s***@gmail.com
2013-10-17 14:37:40 UTC
Permalink
"For those who don't know about it: fossil supports a -M (big emm) option
which reads the commit message from a file and doesn't not apply any
#-related special handling to the content (or it didn't at the time it was
implemented, and i'm assuming that hasn't changed)"
----- stephan beal
Whoa, I didn't know that?
If this is true, why couldn't fossil collapse that same effect for a manual
commit step?
My commit use model is I always select all and delete all predefined
comments and paste in my own.
Unless I am doing a merge and then I take advantage of the autogenerated
sha1 artifact specified.
But, then I have to remember to delete the dang '#' symbol!! :(

It is more work to specify a message file, since my history/commit
documents are appended and not isolated or new per commit.


On Thu, Oct 17, 2013 at 10:23 AM, <***@gmail.com> wrote:

> I also take exception to this being only a recent phenomena. I have been
> burned several times now, but as you say, this is not a mission critical
> error.
> However, me and my users cherish the timeline view and are confused by
> random omissions of '#'this or '#'that. So I do not feel the comments are
> trivial.
>
> I would really appreciate any of the other suggestions.
> Multi-char lead being less onerous.
>
> #Thanks for fossil!
>
>
> On Thu, Oct 17, 2013 at 10:03 AM, j. van den hoff <
> ***@googlemail.com> wrote:
>
>> On Thu, 17 Oct 2013 15:59:45 +0200, Stephan Beal <***@googlemail.com>
>> wrote:
>>
>> On Thu, Oct 17, 2013 at 3:55 PM, j. van den hoff
>>> <***@googlemail.com>**wrote:
>>>
>>> `FOSSIL:' or even `FOS:' as the unique identifier (which mostly is what
>>>> somebody else already proposed). I'd say that will work better than `#'
>>>> (i.e. accidentally names clash very very unlikely) and sufficiently
>>>> stands
>>>> out on the respective lines due to the capitalization.
>>>>
>>>
>>>
>>> Historical note (i don't think you've been on the list long enough to
>>> have
>>> seen this): here's the summary from some code comments in fossil:
>>>
>>> /*
>>> ** Locate the root directory of the local repository tree. The root
>>> ** directory is found by searching for a file named "_FOSSIL_" or
>>> ".fslckout"
>>> ** that contains a valid repository database.
>>> **
>>> ** For legacy, also look for ".fos". The use of ".fos" is deprecated
>>> ** since "fos" has negative connotations in Hungarian, we are told.
>>> ...
>>>
>>> A user made us aware at some point that fos is an ugly/inappropriate word
>>> in his language (i concur with the code comment that it was Hungarian).
>>>
>>
>> well, there are _towns_ having offending names
>>
>> http://en.wikipedia.org/wiki/**Fucking,_Austria<http://en.wikipedia.org/wiki/Fucking,_Austria>:-)
>>
>> so I presume the hungarian users will probably don't take real offense
>> but anyway "FOSSIL:" would be fine to.
>>
>>
>>
>>>
>>
>> --
>> Using Opera's revolutionary email client: http://www.opera.com/mail/
>> ______________________________**_________________
>> fossil-users mailing list
>> fossil-***@lists.fossil-scm.**org <fossil-***@lists.fossil-scm.org>
>> http://lists.fossil-scm.org:**8080/cgi-bin/mailman/listinfo/**
>> fossil-users<http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users>
>>
>
>
Stephan Beal
2013-10-17 14:49:53 UTC
Permalink
On Thu, Oct 17, 2013 at 4:37 PM, <***@gmail.com> wrote:

> If this is true, why couldn't fossil collapse that same effect for a
> manual commit step?
>

i'm not sure what you mean by that?


> My commit use model is I always select all and delete all predefined
> comments and paste in my own.
>

Mine is always to use -m ... ;).

Unless I am doing a merge and then I take advantage of the autogenerated
> sha1 artifact specified.
> But, then I have to remember to delete the dang '#' symbol!! :(
>
> It is more work to specify a message file, since my history/commit
> documents are appended and not isolated or new per commit.
>

AFAIK, nobody really uses the -M option (probably because it's too tedious
to use for most purposes, maybe unless you're auto-importing lists of
changes from another SCM). It was added back in December 2009, and IIRC it
was because someone on the list proposed it and i thought i'd be
interesting (but then i never used it myself).

As far as the other changes go regarding the # symbol - i'm still waiting
for someone else to volunteer :). As a general rule, i don't like to change
features which have survived the test of time (even if they're not 100%
ideal, so long as their limitations are well understood and not generally
problematic).

--
----- stephan beal
http://wanderinghorse.net/home/stephan/
http://gplus.to/sgbeal
"Since tyranny's the only guaranteed byproduct of those who insist on a
perfect world, freedom will have to do." -- Bigby Wolf
s***@gmail.com
2013-10-17 15:01:12 UTC
Permalink
>
> If this is true, why couldn't fossil collapse that same effect for a
> manual commit step?
>
i'm not sure what you mean by that?

me -> I'm asking why can't I perform a manual commit without comments being
ignored, but not using -M.



On Thu, Oct 17, 2013 at 10:49 AM, Stephan Beal <***@googlemail.com>wrote:

> On Thu, Oct 17, 2013 at 4:37 PM, <***@gmail.com> wrote:
>
>> If this is true, why couldn't fossil collapse that same effect for a
>> manual commit step?
>>
>
> i'm not sure what you mean by that?
>
>
>> My commit use model is I always select all and delete all predefined
>> comments and paste in my own.
>>
>
> Mine is always to use -m ... ;).
>
> Unless I am doing a merge and then I take advantage of the autogenerated
>> sha1 artifact specified.
>> But, then I have to remember to delete the dang '#' symbol!! :(
>>
>> It is more work to specify a message file, since my history/commit
>> documents are appended and not isolated or new per commit.
>>
>
> AFAIK, nobody really uses the -M option (probably because it's too tedious
> to use for most purposes, maybe unless you're auto-importing lists of
> changes from another SCM). It was added back in December 2009, and IIRC it
> was because someone on the list proposed it and i thought i'd be
> interesting (but then i never used it myself).
>
> As far as the other changes go regarding the # symbol - i'm still waiting
> for someone else to volunteer :). As a general rule, i don't like to change
> features which have survived the test of time (even if they're not 100%
> ideal, so long as their limitations are well understood and not generally
> problematic).
>
> --
> ----- stephan beal
> http://wanderinghorse.net/home/stephan/
> http://gplus.to/sgbeal
> "Since tyranny's the only guaranteed byproduct of those who insist on a
> perfect world, freedom will have to do." -- Bigby Wolf
>
> _______________________________________________
> fossil-users mailing list
> fossil-***@lists.fossil-scm.org
> http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
>
>
Stephan Beal
2013-10-17 15:34:15 UTC
Permalink
On Thu, Oct 17, 2013 at 5:01 PM, <***@gmail.com> wrote:

> If this is true, why couldn't fossil collapse that same effect for a
>> manual commit step?
>>
> i'm not sure what you mean by that?
>
> me -> I'm asking why can't I perform a manual commit without comments
> being ignored, but not using -M.
>

So you want a manual commit _with_ comments? Sorry, the quadruple negation
here is confusing me: can't... without ... ignored ... not using.

If you want to use '#' in your commit message, either use -M or type them
in using -m:

fossil commit -m '# part 1
# part 2
# part 3' file file2 file3


--
----- stephan beal
http://wanderinghorse.net/home/stephan/
http://gplus.to/sgbeal
"Since tyranny's the only guaranteed byproduct of those who insist on a
perfect world, freedom will have to do." -- Bigby Wolf
Ron Wilson
2013-10-17 15:46:42 UTC
Permalink
On Thu, Oct 17, 2013 at 11:01 AM, <***@gmail.com> wrote:

> I'm asking why can't I perform a manual commit without comments being
> ignored, but not using -M.
>

Maybe a command option or a config option.

The underlying idea of providing and ignoring instructions to the user is
still sound and is in line with Fossil's philosophy of making it easy to
use out-of-the-box.

Also, for what it's worth, my team currently relies on Fossil doing this.
Yes, sometimes ignoring lines starting with # is inconvenient, but it's no
big deal to "hide" the #.

Again, for what it's worth, when we were still using MS Windows (because of
tools being tied to MS Windows), we used CodeWright as our IDE, which had
good support for command line tools. We configured the "Check In" command
as "fossil addremove; fossil commit -M %Q:Enter brief commit message%". For
this, CW would pop-up a text entry dialog with the title "Enter brief
commit message", Then CW would would create a temporary file with the the
entered text and run the command, replacing the %Q:...% with the file name.

Once libfossil is far enough along, we might consider making plug-ins for
the IDEs we currently use.
Stephan Beal
2013-10-17 16:01:39 UTC
Permalink
On Thu, Oct 17, 2013 at 5:46 PM, Ron Wilson <***@gmail.com> wrote:

> Once libfossil is far enough along, we might consider making plug-ins for
> the IDEs we currently use.
>

sidebar/status update: i'm currently in one of my slow phases. i tend to
develop in bursts of 3-5 months or so and then run out of steam for a
couple months. If history repeats itself as predicted, i'll get "back into
it" full speed around Christmas time (where i otherwise have little or
nothing to do because Germany basically shuts down from the 3rd week of
December until the 2nd week of January).

--
----- stephan beal
http://wanderinghorse.net/home/stephan/
http://gplus.to/sgbeal
"Since tyranny's the only guaranteed byproduct of those who insist on a
perfect world, freedom will have to do." -- Bigby Wolf
Andy Bradford
2013-10-17 16:06:14 UTC
Permalink
Thus said Stephan Beal on Thu, 17 Oct 2013 16:49:53 +0200:

> As far as the other changes go regarding the # symbol - i'm still
> waiting for someone else to volunteer :). As a general rule, i don't
> like to change features which have survived the test of time (even if
> they're not 100% ideal, so long as their limitations are well
> understood and not generally problematic).

I don't think changes are required:

$ f ci -M - file <<EOF
> This is a test
> #this
> EOF
New_Version: a037b79c0b552ac0a2c60b6530f19ae5f4022155
$ f stat
repository: /tmp/new/../new.fossil
local-root: /tmp/new/
config-db: /home/amb/.fossil
checkout: a037b79c0b552ac0a2c60b6530f19ae5f4022155 2013-10-17 16:04:45 UTC
parent: 950f43a6f0812dfb7b2a9b894b8bccb1aa056585 2013-10-17 16:04:20 UTC
tags: trunk
comment: This is a test #this (user: amb)

Notice that it preserved my comment.

Fossil FTW!

Andy
--
TAI64 timestamp: 4000000052600b14
s***@gmail.com
2013-10-17 17:06:02 UTC
Permalink
Understood, but the point was to not create a message file.
However, seeing "fossil commit -M fossilcommit.txt" retains '#' lines, I
will just adopt this approach with a script. I actually like this method
more after trying it out. :)

Thanks for the feedback and fossil!


On Thu, Oct 17, 2013 at 12:06 PM, Andy Bradford <amb-***@bradfords.org>wrote:

> Thus said Stephan Beal on Thu, 17 Oct 2013 16:49:53 +0200:
>
> > As far as the other changes go regarding the # symbol - i'm still
> > waiting for someone else to volunteer :). As a general rule, i don't
> > like to change features which have survived the test of time (even if
> > they're not 100% ideal, so long as their limitations are well
> > understood and not generally problematic).
>
> I don't think changes are required:
>
> $ f ci -M - file <<EOF
> > This is a test
> > #this
> > EOF
> New_Version: a037b79c0b552ac0a2c60b6530f19ae5f4022155
> $ f stat
> repository: /tmp/new/../new.fossil
> local-root: /tmp/new/
> config-db: /home/amb/.fossil
> checkout: a037b79c0b552ac0a2c60b6530f19ae5f4022155 2013-10-17 16:04:45
> UTC
> parent: 950f43a6f0812dfb7b2a9b894b8bccb1aa056585 2013-10-17 16:04:20
> UTC
> tags: trunk
> comment: This is a test #this (user: amb)
>
> Notice that it preserved my comment.
>
> Fossil FTW!
>
> Andy
> --
> TAI64 timestamp: 4000000052600b14
> _______________________________________________
> fossil-users mailing list
> fossil-***@lists.fossil-scm.org
> http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
>
Andy Bradford
2013-10-17 17:29:21 UTC
Permalink
Thus said ***@gmail.com on Thu, 17 Oct 2013 13:06:02 -0400:

> Understood, but the point was to not create a message file.

My example does not create a message file but instead tells fossil to
read the commit message from stdin (using a here document):

> > $ f ci -M - file <<EOF
> > > This is a test
> > > #this
> > > EOF
> > New_Version: a037b79c0b552ac0a2c60b6530f19ae5f4022155

No editor is used in this case (except the command line editor) so it
isn't exactly as interactive as using your favorite EDITOR but you don't
have to supply a filename.

Probably not exactly what you were looking for though.

Andy
--
TAI64 timestamp: 4000000052601e90
Ron Wilson
2013-10-17 18:25:11 UTC
Permalink
On Thu, Oct 17, 2013 at 1:29 PM, Andy Bradford <amb-***@bradfords.org>wrote:

> No editor is used in this case (except the command line editor) so it
> isn't exactly as interactive as using your favorite EDITOR but you don't
> have to supply a filename.
>
> Probably not exactly what you were looking for though.
>

How about:

$ cat fci
#! /usr/bin/bash
echo "" > tmp$$.txt
$EDITOR tmp$$.txt
fossil ci -M tmp$$.txt $*
rm tmp$$.txt
$

This will create a temporary file using the process number of the running
script, then delete it after the commit is done.

Optionally, you can pre-load the file with a message template, either as
the parameter the echo or use cp to copy a template file.
Stephan Beal
2013-10-17 18:28:34 UTC
Permalink
On Thu, Oct 17, 2013 at 8:25 PM, Ron Wilson <***@gmail.com> wrote:

> $ cat fci
> #! /usr/bin/bash
> echo "" > tmp$$.txt
> $EDITOR tmp$$.txt
> fossil ci -M tmp$$.txt $*
> rm tmp$$.txt
>

If you want to go one further you can protect against ctrl-c leaving a temp
file laying around by adding:

trap "rm -f tmp$$.txt" 0

somewhere near the top (e.g., right after the first echo "" bit).

OTOH, you might want to keep the temp file if the user Ctrl-C's out in an
untimeline fashion.


> This will create a temporary file using the process number of the running
> script, then delete it after the commit is done.
>

With the trap in place it's removed regardless of how the script exits.

--
----- stephan beal
http://wanderinghorse.net/home/stephan/
http://gplus.to/sgbeal
"Since tyranny's the only guaranteed byproduct of those who insist on a
perfect world, freedom will have to do." -- Bigby Wolf
Ron Wilson
2013-10-16 18:49:00 UTC
Permalink
On Wed, Oct 16, 2013 at 12:57 PM, <***@gmail.com> wrote:

> Yeah, I only asked for a user defined comment character since I thought it
> too bold to request it be changed outright. '#' is really too useful to me
> to be relegated to comments :(
> So many other useless characters to choose from?
>

Besides the various command shells, several languages also use # to
introduce comments.

Note that in commit messages, Fossil only ignore lines that start with #,
so you can "hide" # with other characters.
Loading...