groups.google.com indentation bugs [semi-OT]

K

Keith Thompson

This has been mentioned in several threads. I thought I should post
it in a thread of its own so readers don't miss it unless they want to.

groups.google.com has recently changed its Usenet posting interface.
The new interface has a number of serious bugs that have been
adversely affecting articles posted to comp.lang.c. In many cases, it
appears that the poster is doing something dumb; in fact, it's not the
poster's fault (though there are workarounds in some cases).

The problems I've seen are:

1. Indentation:

Google quietly deletes leading whitespace in most articles. That
includes both tabs and spaces. For example, if I tried to post the
following code through groups.google.com:

int func(void)
{
if (blah)
return 42;
else
return 37;
}

it would probably end up looking like this:

int func(void)
{
if (blah)
return 42;
else
return 37;
}

I've seen it retain some indentation in some cases, but I don't know
what the pattern might be.

A workaround is to precede each line with some non-blank character.
Unfortunately, this makes the code look like quoted text, and makes
it more difficult to cut-and-paste. For example:

.. int func(void)
.. {
.. if (blah)
.. return 42;
.. else
.. return 37;
.. }

Note that this affects only articles posted through groups.google.com,
not articles viewed through groups.google.com.

2. Blank lines:

I've done test posts through groups.google.com in which multiple blank
lines are deleted. That shouldn't be too much of a problem here, but
it's still rude.

3. Hiding of e-mail addresses:

groups.google.com attempts to hide e-mail addresses in displayed
articles, presumably as an anti-spam measure. This affects all
articles viewed through groups.google.com, not just ones posted there.
Apart from this being a questionable idea in the first place (if I
wanted to hide this e-mail address I wouldn't put it in my
signature!), the algorithm apparently messes up any text containing
'@' characters. That's not too much of a problem here, but it's a
real issue for TeX source code and the folks in
rec.games.roguelike.development.

4. Reply interface:

If you click on the "Reply" link while viewing an article, you get a
tiny text box that doesn't quote the article or provide an
attribution. If you click on "Show options" and then on the *other*
"Reply" button, you get a larger text box that properly quotes the
article and automatically provides an attribution line. The latter is
easy to miss, encouraging users to post followups without proper
context.

Note that articles posted through groups.google.com have the
following header:
Organization: http://groups.google.com

I've complained to Google about all these problems. It probably won't
hurt for more people to do so. The support address is
(e-mail address removed). For those of you viewing this though
groups.google.com, that's "groups-support" "at" "google" "dot" "com"
(and my address is "kst-u" "at" "mib" "dot" "org").
 
R

Raymond Martineau

The problems I've seen are:

1. Indentation:
[...]

A workaround is to precede each line with some non-blank character.
Unfortunately, this makes the code look like quoted text, and makes
it more difficult to cut-and-paste. For example:

. int func(void)
. {
. if (blah)
. return 42;
. else
. return 37;
. }

The cut-n-paste problem is solved by the following, but it does take up a
bit more bandwidth (which may be a problem on Usenet, since an article is
copied many times over.)

/**/ int func(void)
/**/ {
/**/ if (blah)
/**/ return 42;
/**/ else
/**/ return 37;
/**/ }

It could also be considered ugly, especially if it is used for a large
number of lines.

3. Hiding of e-mail addresses:

groups.google.com attempts to hide e-mail addresses in displayed
articles, presumably as an anti-spam measure. This affects all
articles viewed through groups.google.com, not just ones posted there.
Apart from this being a questionable idea in the first place (if I
wanted to hide this e-mail address I wouldn't put it in my
signature!), the algorithm apparently messes up any text containing
'@' characters. That's not too much of a problem here, but it's a
real issue for TeX source code and the folks in
rec.games.roguelike.development.

It's probably because of a "public archive" thing - Google attempts to
supress e-mail addresses because the archives could easily be used as a
source to collect e-mail addresses.

I wouldn't see much purpose in that, since an e-mail address that gets
posted to the usenet is can be collected un-altered by spammers and held
for any amount of time.
 
M

Mysidia

Raymond said:
The cut-n-paste problem is solved by the following, but it does take up a
bit more bandwidth (which may be a problem on Usenet, since an article is
copied many times over.)

/**/ int func(void)
/**/ {
/**/ if (blah)
/**/ return 42;
/**/ else
/**/ return 37;
/**/ }

How about this then:

#define _

_ int func(void)
_ {
_ if (blah)
_ return 42;
_ else
_ return 37;
_ }


Then the overhead is once only, is still legitimat code, and the
junk is still easily removed with a regexp

:%s/^_//
 
K

Keith Thompson

Mysidia said:
How about this then:

#define _

_ int func(void)
_ {
_ if (blah)
_ return 42;
_ else
_ return 37;
_ }


Then the overhead is once only, is still legitimat code, and the
junk is still easily removed with a regexp

:%s/^_//

That doesn't work if the "_" is on a line with a preprocessor
directive:

#define _
_ #include <stdio.h> /* syntax error */
_ int main(void)
_ {
_ printf("This doesn't work\n");
_ return 0;
_ }

It also means that the code posted isn't the same as the code you
actually compiled, which has always been a major source of confusion.

It turns out that there's a much better workaround: use
groups.google.ca, which still uses the old working interface, rather
than groups.google.com, which is broken.

The best solution, of course, is for Google to fix their stupid
interface, and revert to the old one until the bugs are fixed.
 
O

Old Wolf

Keith said:
It turns out that there's a much better workaround: use
groups.google.ca, which still uses the old working interface, rather
than groups.google.com, which is broken.

You can't actually post properly using the old interface. When
you try to reply to a thread, you get in red:

.. Unable to retrieve message woxEd.76382$dv1.70730@edtnps89

and no quoted text appears. I haven't actually tried this, but I
presume that manually quoting the text and posting will start
a new thread (since it couldn't find the reference to the old
thread to put in the message).

Also, the old interface is often up to 12 hours behind the
latest posts.
 
T

Tomi Häsä

Keith Thompson said:
groups.google.com has recently changed its Usenet posting interface.
The new interface has a number of serious bugs that have been
adversely affecting articles posted to comp.lang.c. In many cases, it
appears that the poster is doing something dumb; in fact, it's not the
poster's fault (though there are workarounds in some cases).

The problems I've seen are:

1. Indentation:

Google quietly deletes leading whitespace in most articles. That
includes both tabs and spaces. For example, if I tried to post the
following code through groups.google.com:

int func(void)
{
if (blah)
return 42;
else
return 37;
}

it would probably end up looking like this:

int func(void)
{
if (blah)
return 42;
else
return 37;
}

I've seen it retain some indentation in some cases, but I don't know
what the pattern might be.

I have studied the indentation or actually more about changing between
fixed-width and variable-width font here:

http://groups-beta.google.com/group...c0e63a72884/efa81dce01fc5753#efa81dce01fc5753

Here are some other problems I have found with GG2:

http://groups-beta.google.com/group/alt.fan.dejanews/msg/27cf986c15a7c813
3. Hiding of e-mail addresses:

groups.google.com attempts to hide e-mail addresses in displayed
articles, presumably as an anti-spam measure. This affects all
articles viewed through groups.google.com, not just ones posted there.
Apart from this being a questionable idea in the first place (if I
wanted to hide this e-mail address I wouldn't put it in my
signature!), the algorithm apparently messes up any text containing
'@' characters. That's not too much of a problem here, but it's a
real issue for TeX source code and the folks in
rec.games.roguelike.development.

Here's a study about the munging of characters near @ character:

http://groups-beta.google.com/group...99eab27b0c0/755dbeccf7668bb1#755dbeccf7668bb1
Note that articles posted through groups.google.com have the
following header:
Organization: http://groups.google.com

The articles also have this header, which isn't added to the messages posted
with GG1:

User-Agent: G2/0.2
I've complained to Google about all these problems. It probably won't
hurt for more people to do so. The support address is
(e-mail address removed). For those of you viewing this though
groups.google.com, that's "groups-support" "at" "google" "dot" "com"
(and my address is "kst-u" "at" "mib" "dot" "org").

And you can also report problems with GG2 in this official Google Groups Beta
discussion forum:

http://groups-beta.google.com/group/google-labs-groups2
 
T

Tomi Häsä

T

Taladan

Tomi said:
http://groups-beta.google.com/group...c0e63a72884/efa81dce01fc5753#efa81dce01fc5753

And here's another test, which shows that when there are four or more special
characters on a line, the line will be shown using monospace font:

http://groups-beta.google.com/group...7b27d1534cb/01cbeccef829695e#01cbeccef829695e


Here's another workaround:


http://www.mozilla.org/products/thunderbird/


built in newsgroup browser. It's what I decided to go with after I
realized that google groups borks code statements. Just an idea
(thought I don't know if it'll be kosher if you're @ work, depends on
your boss/firm).


Tal
 
K

Keith Thompson

Taladan said:

Yes, that's one of a number of Usenet client programs. They all
require access to an NNTP server. Your ISP may provide one, or you
can use a free server such as news.individual.net.

One bit of bad AOL has announced that they're about to drop
Usenet support. They're recommending that their users switch to
groups.google.com.
 
F

Flash Gordon

Chris said:
That's good news! September is nearly over!

Only in terms of one group.
/That's/ the bad news...

Perhaps someone should post all over AOL support groups that
news.individual.net a number of other officially free news servers are
better.
 

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments. After that, you can post your question and our members will help you out.

Ask a Question

Members online

Forum statistics

Threads
473,767
Messages
2,569,572
Members
45,045
Latest member
DRCM

Latest Threads

Top