This time my problem is that a long commit message does not appear as part of the patch.

Here's what I did:

1 darcs record
2 I said "yes" to the "Do you want to add a long comment?" question
3 Darcs popped up an editor window for a file darcs-record-0
4 I put in the (long) comment
5 I closed the editor

But the comment is not recorded.  Because of this, I tried a second time, changing one comment line in the key file.  This time, when it popped up darcs-record-0, my long comment was there already.  So I saved (a no-op) and closed the editor.  Again, the message is not recorded as part of the patch.

I can't explain this.  I've previously managed to get messages into patches.  The only odd thing about this one is that it's quite long (a couple of pages).

Now I'm stumped.

I hope you don't mind my stream of reports.  I'm a rather naïve darcs user -- but I'm faithfully taking time to report problems as I come across them, because I'm assuming you *want* to hear from naïve users.

Simon

On Mon, Nov 13, 2006 at 12:46:24PM +0000, simonpj wrote:
> I hope you don't mind my stream of reports.  I'm a rather naive darcs
> user -- but I'm faithfully taking time to report problems as I come
> across them, because I'm assuming you *want* to hear from naive users.

I agree with Juliusz here, although I suspect I'm even more busy than he is
(it being my first term teaching).  We definitely welcome reports, although
we also wish you were having fewer problems.
-- 
David Roundy
Dept. of Physics
Oregon State University

> Here's what I did:
> 
> 1 darcs record
> 2 I said "yes" to the "Do you want to add a long comment?" question
> 3 Darcs popped up an editor window for a file darcs-record-0
> 4 I put in the (long) comment
> 5 I closed the editor
> 
> But the comment is not recorded.  Because of this, I tried a second
> time, changing one comment line in the key file.  This time, when it
> popped up darcs-record-0, my long comment was there already.  So I
> saved (a no-op) and closed the editor.  Again, the message is not
> recorded as part of the patch.

If you try to record the same patch again, but with a short log message,
does it go through?  Similarly, if you create a new repository, add a
file or something, and record it with the same commit message that is
giving you problems, what happens?  How big is the file (filesize and
lines)?

Can you check the permissions on the darcs-record-0 file?  I doubt this
is the issue, but if it's a quick check I'd like to get that out of the
way.

[snip]

> I hope you don't mind my stream of reports.  I'm a rather na??ve darcs
> user -- but I'm faithfully taking time to report problems as I come
> across them, because I'm assuming you *want* to hear from na??ve
> users.

I certainly appreciate anyone who takes the time to fill out bug
reports.  Even if it is caused by their lack of understanding about the
application, they still took the time to try to help.  Besides, if users
are getting confused by things, that could be a UI bug of sorts :P

Just to eliminate an obvious possibility, I recorded a patch with a bible as
long comment (100,000+ lines), without problems ... well, it wasn't particularly
rapid. Both 1.0.8 and unstable.

(Better say I like bug reports too, don't want to be the oddball ;-)

I have not heard from anyone about the problem below, nor has the bug (if it is one) appeared in the bug tracker.
Currently, Darcs on Windows is not recording *any* long comments. I type them in, and they simply get discarded.  It's very tiresome.

Furthermore, once recorded, I cannot un-record (with a view to re-recording), not even in the *very next command*.  This latter problem is Issue334, but again I got no feedback, so I'm still stumped.

Frustrated,

Simon

Hi Simon,

On Thu, Nov 23, 2006 at 11:48:14 +0000, simonpj wrote:
> I have not heard from anyone about the problem below, nor has the bug
> (if it is one) appeared in the bug tracker.

Did you mean: http://bugs.darcs.net/issue345 
or something else?

It is troubling if you're not getting any of the bugtracker traffic on
the issue.  Interestingly, Juliusz's reply is not being tracked either.

> Currently, Darcs on Windows is not recording *any* long comments. I
> type them in, and they simply get discarded.  It's very tiresome.

Well, there is is the problem of no-Windows.  I'm home with my parents
for Thanksgiving and can try this out on their machine.  Hopefully, my
WinXP copy will have arrived when I get back in France and this kind
of stuff will get a little easier.

But in the meantime, we are all very happy to have your bug reports and
unhappy to have the bugs behind them.

| Well, there is is the problem of no-Windows.  I'm home with my parents
| for Thanksgiving and can try this out on their machine.  Hopefully, my
| WinXP copy will have arrived when I get back in France and this kind
| of stuff will get a little easier.

Eric, I posted you a copy of Windows XP Pro several weeks ago.  I sent you a message when I did so (enclosed).  Didn't you get it?

Simon

Attachments

• unnamed

On Thu, Nov 23, 2006 at 14:09:27 +0000, Simon Peyton-Jones wrote:
> Eric, I posted you a copy of Windows XP Pro several weeks ago.  I sent
> you a message when I did so (enclosed).  Didn't you get it?

I did get your message (sorry, I was trying to be mail-efficient by only
responding when I received it), but it had not arrived when I left for
the States, and the lab have not notified me of any packages arriving
for me.  Hmm... I'll go check with my colleagues to see if they have
anything.

Just a couple of questions here to help me reproduce the issue:

1. What environment is this (MSYS, Cygwin, DOS prompt, other)?
2. What is EDITOR (or DARCS_EDITOR) set to?
3. If it's a Unixy text editor, how was it built? (MSYS, Cygwin, other)

Thanks,

I'm using MSYS.  I invoke darcs from a bash shell running inside emacs.
The editor is emacs
It's just the standard Gnu emacs windows distribution.  I did not build it.

I have tried in a fresh tree.  It works.  Maybe it's something to do with my particular tree. Shall I tar it up for you?

Simon

| -----Original Message-----
| From: Eric Kow [mailto:bugs@darcs.net]
| Sent: 23 November 2006 14:27
| To: droundy@darcs.net; eric.kow@gmail.com; kapheine@divineinvasion.net; ptp@lysator.liu.se; Simon
| Peyton-Jones
| Subject: [issue345] Commit message is not recorded
|
|
| Eric Kow <eric.kow@gmail.com> added the comment:
|
| Just a couple of questions here to help me reproduce the issue:
|
| 1. What environment is this (MSYS, Cygwin, DOS prompt, other)?
| 2. What is EDITOR (or DARCS_EDITOR) set to?
| 3. If it's a Unixy text editor, how was it built? (MSYS, Cygwin, other)
|
| Thanks,
|
| ____________________________________
| Darcs issue tracker <bugs@darcs.net>
| <http://bugs.darcs.net/issue345>
| ____________________________________

On Thu, Nov 23, 2006 at 17:00:05 +0000, Simon Peyton-Jones wrote:
> I have tried in a fresh tree.  It works.  Maybe it's something to do
> with my particular tree. Shall I tar it up for you?

Yes, thanks.

http://research.microsoft.com/~simonpj/tmp/darcs-bug-24Nov06.tar.gz

| -----Original Message-----
| From: Eric Kow [mailto:bugs@darcs.net]
| Sent: 23 November 2006 19:16
| To: droundy@darcs.net; eric.kow@gmail.com; kapheine@divineinvasion.net; ptp@lysator.liu.se; Simon
| Peyton-Jones
| Subject: [issue345] Commit message is not recorded
|
|
| Eric Kow <eric.kow@gmail.com> added the comment:
|
| On Thu, Nov 23, 2006 at 17:00:05 +0000, Simon Peyton-Jones wrote:
| > I have tried in a fresh tree.  It works.  Maybe it's something to do
| > with my particular tree. Shall I tar it up for you?
|
| Yes, thanks.
|
| ____________________________________
| Darcs issue tracker <bugs@darcs.net>
| <http://bugs.darcs.net/issue345>
| ____________________________________

On Fri, Nov 24, 2006 at 10:31:59 +0000, Simon Peyton-Jones wrote:
> http://research.microsoft.com/~simonpj/tmp/darcs-bug-24Nov06.tar.gz

Hmm... the _darcs seems to be missing from this.
Is that normal?

Stupid me.  I've uploaded one that has _darcs.  (I was trying to avoid including libraries/ and dropped too much)

it'll take a few mins to upload
S

| -----Original Message-----
| From: Eric Y. Kow [mailto:eric.kow@gmail.com]
| Sent: 24 November 2006 12:40
| To: Simon Peyton-Jones
| Cc: Darcs issue tracker; droundy@darcs.net; kapheine@divineinvasion.net; ptp@lysator.liu.se
| Subject: Re: [issue345] Commit message is not recorded
|
| On Fri, Nov 24, 2006 at 10:31:59 +0000, Simon Peyton-Jones wrote:
| > http://research.microsoft.com/~simonpj/tmp/darcs-bug-24Nov06.tar.gz
|
| Hmm... the _darcs seems to be missing from this.
| Is that normal?
|
| --
| Eric Kow                     http://www.loria.fr/~kow
| PGP Key ID: 08AC04F9         Merci de corriger mon français.

On Fri, Nov 24, 2006 at 12:47:19 +0000, Simon Peyton-Jones wrote:
> Stupid me.  I've uploaded one that has _darcs.  (I was trying to avoid including libraries/ and dropped too much)

So I've had a look at the repository, using the standard GNU Emacs with
the MSYS shell running via M-x shell.  DARCS_EDITOR is set to emacs.exe
with the emacs bin stuff in my PATH.

I tried a darcs record, first with only one change, and then with all
the changes and each time I've been able to get long messages in.
So I'm a bit stumped.  I guess since you're up and running, this isn't
very urgent, but it is very weird to me.

can you unrecord too?   That is indeed strange.  As you say we may have to leave it if you can't reproduce it.

S

| -----Original Message-----
| From: Eric Y. Kow [mailto:eric.kow@gmail.com]
| Sent: 24 November 2006 19:35
| To: Simon Peyton-Jones
| Cc: Darcs issue tracker; droundy@darcs.net; kapheine@divineinvasion.net; ptp@lysator.liu.se
| Subject: Re: [issue345] Commit message is not recorded
|
| On Fri, Nov 24, 2006 at 12:47:19 +0000, Simon Peyton-Jones wrote:
| > Stupid me.  I've uploaded one that has _darcs.  (I was trying to avoid including libraries/ and dropped
| too much)
|
| So I've had a look at the repository, using the standard GNU Emacs with
| the MSYS shell running via M-x shell.  DARCS_EDITOR is set to emacs.exe
| with the emacs bin stuff in my PATH.
|
| I tried a darcs record, first with only one change, and then with all
| the changes and each time I've been able to get long messages in.
| So I'm a bit stumped.  I guess since you're up and running, this isn't
| very urgent, but it is very weird to me.
|
| --
| Eric Kow                     http://www.loria.fr/~kow
| PGP Key ID: 08AC04F9         Merci de corriger mon français.

[Resending because of silly typo that bugtracker rejected]

On Mon, Nov 27, 2006 at 08:05:17 +0000, Simon Peyton-Jones wrote:
> can you unrecord too?   That is indeed strange.  As you say we may have to leave it if you can't reproduce it.

Sorry about not responding sooner.

Speaking from memory (I'm no longer on that machine), I have tried
unrecording the patch and then recording again.  Seems to work.
So, no luck reproducing this :-(

I'm going to mark this need-eg until somebody else has a similar,
perhaps easier-to-reproduce bug (or an idea what might be causing
this)

With no further follow-ups and no "me too" comments in over a year, I'm marking
this as wont-fix for now. (More like "Can't fix"). 

Certainly if anyone is aware of the issue still being present, please re-open
the ticket. It sounds annoying to live with.

Actually, I think this falls more in the category of a cant-reproduce bug.  It
looks like Simon had no trouble in another repository.  I'm not quite sure how
best to categorize bugs that are unreproducible.  wont-fix seems wrong, and
need-eg seems right, but not quite strong enough.  Perhaps we should introduce
some sort of a "presumed-dead" status for bugs?

David

David Roundy wrote:
> 
> Actually, I think this falls more in the category of a cant-reproduce bug.  It
> looks like Simon had no trouble in another repository.  I'm not quite sure how
> best to categorize bugs that are unreproducible.  wont-fix seems wrong, and
> need-eg seems right, but not quite strong enough.  Perhaps we should introduce
> some sort of a "presumed-dead" status for bugs?

I generally agree with you, David.

We possibly *would* fix it if we could reproduce it, but it's not just 
that we waiting on an example....there is no example to be easily 
produced. "presumed-dead" is OK with me.

    Mark