Many applications give unhelpful error messages, but I've never met any quite as useless as the invariable message from BackupPC: "Unable to read 4 bytes" If only it could sometimes say "Unable to read 3 bytes" or even "Hooray, was able to read 1 byte".
What 4 bytes did it want to read, incidentally?
On Fri, 2014-06-20 at 15:50 +0100, Timothy Murphy wrote:
Many applications give unhelpful error messages, but I've never met any quite as useless as the invariable message from BackupPC: "Unable to read 4 bytes" If only it could sometimes say "Unable to read 3 bytes" or even "Hooray, was able to read 1 byte".
What 4 bytes did it want to read, incidentally?
AFAIK it's actually an error from rsync (I've seen it in rsnapshot as well, which is also rsync-based). Google for more.
poc
On Fri, 20 Jun 2014 16:48:38 +0100 Patrick O'Callaghan wrote:
AFAIK it's actually an error from rsync (I've seen it in rsnapshot as well, which is also rsync-based). Google for more.
And apparently, what it actually means is: "Hey! I'm trying to talk to an incompatible version of rsync on the other system."
I got it recently trying to copy a very old system's root to a virtual image and I had to download the source rpm for rsync and build the new rsync on the old system.
On Fri, 2014-06-20 at 12:13 -0400, Tom Horsley wrote:
On Fri, 20 Jun 2014 16:48:38 +0100 Patrick O'Callaghan wrote:
AFAIK it's actually an error from rsync (I've seen it in rsnapshot as well, which is also rsync-based). Google for more.
And apparently, what it actually means is: "Hey! I'm trying to talk to an incompatible version of rsync on the other system."
I got it recently trying to copy a very old system's root to a virtual image and I had to download the source rpm for rsync and build the new rsync on the old system.
Yes, that sounds about right. I was too lazy to backtrack and look it up again :-)
poc
On Fri, 20 Jun 2014 19:16:29 +0100, "Patrick O'Callaghan" pocallaghan@gmail.com wrote:
On Fri, 2014-06-20 at 12:13 -0400, Tom Horsley wrote:
On Fri, 20 Jun 2014 16:48:38 +0100 Patrick O'Callaghan wrote:
AFAIK it's actually an error from rsync (I've seen it in rsnapshot as well, which is also rsync-based). Google for more.
And apparently, what it actually means is: "Hey! I'm trying to talk to an incompatible version of rsync on the other system."
I got it recently trying to copy a very old system's root to a virtual image and I had to download the source rpm for rsync and build the new rsync on the old system.
Yes, that sounds about right. I was too lazy to backtrack and look it up again :-)
poc
It can also occur with BackupPC when keyless ssh isn't set up right and ssh is giving a prompt (e.g., on the first ssh login to a host). Has happened to me more than once...
George
On Fri, 20 Jun 2014 14:21:52 -0400 George Avrunin wrote:
It can also occur with BackupPC when keyless ssh isn't set up right and ssh is giving a prompt (e.g., on the first ssh login to a host). Has happened to me more than once...
Yea, that's the reason you find everywhere when googling, which is why it took me forever to figure out the version incompatibility that gives the same error :-).
Patrick O'Callaghan wrote:
What 4 bytes did it want to read, incidentally?
AFAIK it's actually an error from rsync (I've seen it in rsnapshot as well, which is also rsync-based). Google for more.
Do you mean that rsync actually sent the message, or that it was an error in rsync that motivated BackupPC to send it?
I've never had the message directly from rsync, although I use it all the time.
Also, I'm pretty sure (but not certain) that I've had the message from BackupPC for other reasons, perhaps before rsync is called.
In any case, BackupPC must know that rsync has failed, so it would be helpful if it added "Rsync failure" or some such. I found recently that actually running the rsync command specified in BackupPC (in Edit Config, on the web page) the cause of the problem immediately became clear.
Incidentally, in defence of my rant, I've found that when I actually try to trace an error message in the source - I used to do this quite a lot with NetworkManager when it used to send out error messages which just said "Failure 27" or something like that - I usually found that the cause was more or less obvious.
I wish developers would spend a moment when adding error messages, thinking "Will this message actually help someone encountering this error to correct the problem?" As an example of a very rare program which does this well I'm mention SELinux, which always suggests what one probably needs to do.
Allegedly, on or about 22 June 2014, Timothy Murphy sent:
Incidentally, in defence of my rant, I've found that when I actually try to trace an error message in the source - I used to do this quite a lot with NetworkManager when it used to send out error messages which just said "Failure 27" or something like that - I usually found that the cause was more or less obvious.
I wish developers would spend a moment when adding error messages, thinking "Will this message actually help someone encountering this error to correct the problem?"
I was similarly pissed off when trying to help a friend with his Windows PC. He needed to activate it (another thing I hate), but it kept failing with some obtuse 8-digit error code. I had to use another computer to google that code (fortunately, I can do that here, but he wouldn't have been able to at his home). It turns that that said stupid error code was due to the clock needing to be set correctly, first. Of all the crappy stupid error messages that mickysoft has ever inflicted upon users, that really takes the cake. Why the hell couldn't the computer just have said to set the clock manually, since it couldn't manage to do it itself, instead of some moronic number code? And finding how to manually set a the clock on Windows 8 was no fun, either.
Sodding computers, they're designed to send you mad, I'm sure.
On 22 June 2014 17:04, Tim ignored_mailbox@yahoo.com.au wrote:
Why the hell couldn't the computer just have said to set the clock manually, since it couldn't manage to do it itself, instead of some moronic number code?
I know this was a rhetorical question, but:
These days software is expected to be localised into hundreds of different languages. Until that is done, the programmer does not know and must not assume what script/alphabet/language/character set an error might appear in, let alone what words it will contain. It could be a couple of characters of an ideographic script such as Chinese, which can convey a whole sentence in 2 or 3 "letters", or it could be several lines of hundreds of characters in the case of some Polynesian languages with restricted phonemic inventories. (I pick 2 extreme cases to make a point.)
And errors by nature are unexpected. If, say, the system language has not yet been set at this point, then the right thing to do is not to display a message that the user might not be able to read or even copy down - could you write down a message in Japanese? And type it into Google? I speak a little Japanese and I couldn't - but to show a simple number that they could.
Tim:
Why the hell couldn't the computer just have said to set the clock manually, since it couldn't manage to do it itself, instead of some moronic number code?
Liam Proven:
I know this was a rhetorical question, but:
These days software is expected to be localised into hundreds of different languages. Until that is done, the programmer does not know and must not assume what script/alphabet/language/character set an error might appear in, let alone what words it will contain.
That's all well and true, but you'd expect an American english program to have error messages in that language, at least. Sure, by all means put an easy to copy and paste error code for uses to search for more info it needs it, as well, but put a written message along with it.
It's just the sort of moronic thing that we'd come to expect with Winblows. What could have been an easily user-fixable thing (reset the clock), becomes a service call for average users who don't know how to get past error number x0e4343243 which gives no clue, nor any way to find out what it might mean if you can't do a google search.
On 23 June 2014 20:33, Tim ignored_mailbox@yahoo.com.au wrote:
Tim:
Why the hell couldn't the computer just have said to set the clock manually, since it couldn't manage to do it itself, instead of some moronic number code?
Liam Proven:
I know this was a rhetorical question, but:
These days software is expected to be localised into hundreds of different languages. Until that is done, the programmer does not know and must not assume what script/alphabet/language/character set an error might appear in, let alone what words it will contain.
That's all well and true, but you'd expect an American english program to have error messages in that language, at least. Sure, by all means put an easy to copy and paste error code for uses to search for more info it needs it, as well, but put a written message along with it.
It's just the sort of moronic thing that we'd come to expect with Winblows. What could have been an easily user-fixable thing (reset the clock), becomes a service call for average users who don't know how to get past error number x0e4343243 which gives no clue, nor any way to find out what it might mean if you can't do a google search.
Your quoting is... odd.
But no, not really. If you don't leave space for text, you just raise a numbered error and another program catches that and replaces the number with a message. It's a lot easier to Google a message.
But it's good practice to /avoid/ presenting text errors at the lowest level.
Let's say you have no cut and paste available to you. Could you Google this on another computer?
ファイルプログラム読み込み数のモジュールのデコードでエラーを見つけていない。
Now let's assume we were both Japanese. Could we record and search accurately for a written sentence in English?
Probably not.
But we could write down 795237975 and search for it...
On 23 June 2014 21:16, Joe Zeff joe@zeff.us wrote:
On 06/23/2014 11:43 AM, Liam Proven wrote:
Now let's assume we were both Japanese. Could we record and search accurately for a written sentence in English?
Considering that most Japanese students study English in school? Probably.
:¬) All right, all right!
Substitute $ALPHABET of choice, then...
On 23 June 2014 19:43, Liam Proven lproven@gmail.com wrote:
On 23 June 2014 20:33, Tim ignored_mailbox@yahoo.com.au wrote:
Tim:
Why the hell couldn't the computer just have said to set the clock manually, since it couldn't manage to do it itself, instead of some moronic number code?
Liam Proven:
I know this was a rhetorical question, but:
These days software is expected to be localised into hundreds of different languages. Until that is done, the programmer does not know and must not assume what script/alphabet/language/character set an error might appear in, let alone what words it will contain.
That's all well and true, but you'd expect an American english program to have error messages in that language, at least. Sure, by all means put an easy to copy and paste error code for uses to search for more info it needs it, as well, but put a written message along with it.
It's just the sort of moronic thing that we'd come to expect with Winblows. What could have been an easily user-fixable thing (reset the clock), becomes a service call for average users who don't know how to get past error number x0e4343243 which gives no clue, nor any way to find out what it might mean if you can't do a google search.
Your quoting is... odd.
But no, not really. If you don't leave space for text, you just raise a numbered error and another program catches that and replaces the number with a message. It's a lot easier to Google a message.
But it's good practice to /avoid/ presenting text errors at the lowest level.
Let's say you have no cut and paste available to you. Could you Google this on another computer?
ファイルプログラム読み込み数のモジュールのデコードでエラーを見つけていない。
Now let's assume we were both Japanese. Could we record and search accurately for a written sentence in English?
Probably not.
But we could write down 795237975 and search for it...
The example is slightly misleading. English is the most widely spoken language and latin is the most widely used alphabet. You can't copy kanji or Chinese characters without familiarity with that writing system. You might be able to copy kana, or other phonetic/syllabic systems, if you can recognise them as such, but most people outside Japan are not going to spot that.
It's perfectly feasible in the above example to have returned a number and a localised interpretation.
Though I'll bet the error doesn't actually mean that the clock is not set correctly. The usual experience is it'll be something else going wrong for which the underlying cause is the clock set incorrectly and anonymous error is being reported that no-one ever expected to get to.
Allegedly, on or about 23 June 2014, Ian Malone sent:
Though I'll bet the error doesn't actually mean that the clock is not set correctly. The usual experience is it'll be something else going wrong for which the underlying cause is the clock set incorrectly and anonymous error is being reported that no-one ever expected to get to.
But who could tell? Because the error message is useless. The information mickysoft provides on their website about that error only mentioned that you needed to set your clock. And the plethora of blind-leading-the-blind on internet help sites is completely unreliable when it comes to closed source, and inadequately documented software.
Which gets back to the original poster's point: Software authors should make useful error message reports. Ones that correctly identify the fault, and allow you to take the appropriate actions.
NetworkManager was one example of crap response codes (unexplained, anywhere, numerical codes). And Evolution is another. How many of us have sat there waiting for it to finish dealing with some "unknown background process" before we can continue using it? The galling thing is that /that/ "unknown" process is one of its own.
On 24 June 2014 09:13, Tim ignored_mailbox@yahoo.com.au wrote:
Allegedly, on or about 23 June 2014, Ian Malone sent:
Though I'll bet the error doesn't actually mean that the clock is not set correctly. The usual experience is it'll be something else going wrong for which the underlying cause is the clock set incorrectly and anonymous error is being reported that no-one ever expected to get to.
But who could tell? Because the error message is useless. The information mickysoft provides on their website about that error only mentioned that you needed to set your clock. And the plethora of blind-leading-the-blind on internet help sites is completely unreliable when it comes to closed source, and inadequately documented software.
Which gets back to the original poster's point: Software authors should make useful error message reports. Ones that correctly identify the fault, and allow you to take the appropriate actions.
My point was if the error report existed it might still not have been useful if it wasn't coming directly from some kind of clock check. What happens then is you get people running across this error and once it's figured out they list the code on their website against this problem, then you find it and fix it through the magic of google. Not ideal, especially if you don't have alternative internet access, but most ways of dealing with post-release problems aren't ideal.
PS, you've misspelt Microsoft. If you want to be consistent you need to make up humorous[1] misspellings of NetworkManager and Evolution too. [1] For low values of humorous.
Allegedly, on or about 24 June 2014, Ian Malone sent:
PS, you've misspelt Microsoft. If you want to be consistent you need to make up humorous[1] misspellings of NetworkManager and Evolution too.
They've worked very hard to earn their denigrations...
On 24 June 2014 12:03, Ian Malone ibmalone@gmail.com wrote:
PS, you've misspelt Microsoft. If you want to be consistent you need to make up humorous[1] misspellings of NetworkManager and Evolution too. [1] For low values of humorous.
I suggest MICROS~1. It's what Win9x turned the name of the "Microsoft Office" folder into -- i.e. it's the company's own mangling of its own name.
On 25 June 2014 16:09, Liam Proven lproven@gmail.com wrote:
On 24 June 2014 12:03, Ian Malone ibmalone@gmail.com wrote:
PS, you've misspelt Microsoft. If you want to be consistent you need to make up humorous[1] misspellings of NetworkManager and Evolution too. [1] For low values of humorous.
I suggest MICROS~1. It's what Win9x turned the name of the "Microsoft Office" folder into -- i.e. it's the company's own mangling of its own name.
I think that might infringe some of their patents...
On Sun, 2014-06-22 at 15:24 +0100, Timothy Murphy wrote:
Do you mean that rsync actually sent the message, or that it was an error in rsync that motivated BackupPC to send it?
IIRC when I saw it it was something in /var/log/rsnapshot or /var/log/cron, but it was a while back.
poc
-
George Avrunin -
Ian Malone -
Joe Zeff -
Liam Proven -
Marcus Leech -
Patrick O'Callaghan -
Tim -
Timothy Murphy -
Tom Horsley