Bug#422085: bugscripts are executed, but content not put in email
reopen 422085 thanks Hi! I just tried using reportbug-ng to report bugs against packages like texlive- doc-base, xserver-xorg and mdadm. (what's common for these 3 packages is that thay have bug scripts) And the result is: I see a terminal pop up briefly with a lot of text scrolling over and a line added to the email with the following: --- Output from package bug script --- but no actual output from bug script. I use kmail as the email client - and running reportbug-ng with -l info doesn't reveal any info about why it isn't adding the bug script output. /Sune -- How may I explore the directory? You should never close a icon for doubleclicking on the button. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#422085: closed by Bastian Venthur [EMAIL PROTECTED] (Re: Bug#422085: bugscripts are executed, but content not put in email)
On Mon, Jul 21, 2008 at 09:00:04 +, Debian Bug Tracking System wrote: This is not a bug of rng and not related to 422085: err. yes it is. * calling /usr/share/bug/texlive-doc-base 31 aborts with an error (getkey: command not found) when called in the console. That's one of the documented functions such a script can rely on. If you feel this is a bug of rng, please open a new report specific to this problem and not to rng not using the scripts at all. * xserver-xorg: the output of the script is too long to handle so rng added the following text to the mail (you should have seen that, Sune): -8---8---8---8---8---8---8---8---8-- Please attach the file: /tmp/reportbug-ng-xserver-xorg-oob9Xu.txt to the mail. I'd do it myself if the output wasn't too long to handle. Thank you! -8---8---8---8---8---8---8---8---8-- as you can see rng already gathered the info and even put in in a file, all the user has to do is to attach this file to the mail. That doesn't make any sense. If you gather the info, then surely putting it in the mail is not much more work. I'm pretty sure you can attach files to a mail, if nothing else. * mdadm: not tested, since not installed, but I assume its one of the above problems. Try udev, it also comes with a script whose output isn't 80kb big and it works fine, rng attaches the output automatically and the user has nothing to do. The file too long problem isn't rng's fault but a limitation of the shell which can't execute commands of unlimited length. Rng detects such cases where calling the MUA failed and puts the output in a file and asks the user kindly to attach it. Yes, it's rng's fault. I'm not sure why you say the shell has anything to do with this. Cheers, Julien -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#422085: closed by Bastian Venthur [EMAIL PROTECTED] (Re: Bug#422085: bugscripts are executed, but content not put in email)
Julien Cristau wrote: On Mon, Jul 21, 2008 at 09:00:04 +, Debian Bug Tracking System wrote: * calling /usr/share/bug/texlive-doc-base 31 aborts with an error (getkey: command not found) when called in the console. That's one of the documented functions such a script can rely on. Where is that documented? And where does getkey come from? Haven't found a package providing this command. as you can see rng already gathered the info and even put in in a file, all the user has to do is to attach this file to the mail. That doesn't make any sense. If you gather the info, then surely putting it in the mail is not much more work. I'm pretty sure you can attach files to a mail, if nothing else. Not every MUA supports an attach file option. It's not that I haven't already tried it (#491499). So since attaching the file automatically is not possible right now, asking the user kindly to attach the file is the second best rng can do. And please keep in mind that this only happens when the output of a script is *very* large -- to large for the shell to handle -- like the one from xserver-xorg, for the most other packages, everything should work fine automatically and attaching a file is not needed. The file too long problem isn't rng's fault but a limitation of the shell which can't execute commands of unlimited length. Rng detects such cases where calling the MUA failed and puts the output in a file and asks the user kindly to attach it. Yes, it's rng's fault. I'm not sure why you say the shell has anything to do with this. How do you think rng invokes the different MUAs? If you have a better solution which supports calling MUAs with bodies of arbitrary textlengths, then please send me a patch. I'd love to have this included rather to rely on the user to attach a file. -- Bastian Venthur http://venthur.de Debian Developer venthur at debian org -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#422085: closed by Bastian Venthur [EMAIL PROTECTED] (Re: Bug#422085: bugscripts are executed, but content not put in email)
On Mon, Jul 21, 2008 at 17:15:10 +0200, Bastian Venthur wrote: Julien Cristau wrote: On Mon, Jul 21, 2008 at 09:00:04 +, Debian Bug Tracking System wrote: * calling /usr/share/bug/texlive-doc-base 31 aborts with an error (getkey: command not found) when called in the console. That's one of the documented functions such a script can rely on. Where is that documented? And where does getkey come from? Haven't found a package providing this command. /usr/share/doc/reportbug/README.developers.gz The function is in /usr/share/reportbug/handle_bugscript which looks like it could be used pretty much as-is in rng. as you can see rng already gathered the info and even put in in a file, all the user has to do is to attach this file to the mail. That doesn't make any sense. If you gather the info, then surely putting it in the mail is not much more work. I'm pretty sure you can attach files to a mail, if nothing else. Not every MUA supports an attach file option. Then they're broken. It's not that I haven't already tried it (#491499). So since attaching the file automatically is not possible right now, asking the user kindly to attach the file is the second best rng can do. And please keep in mind that this only happens when the output of a script is *very* large -- to large for the shell to handle -- like the one from xserver-xorg, for the most other packages, everything should work fine automatically and attaching a file is not needed. The file too long problem isn't rng's fault but a limitation of the shell which can't execute commands of unlimited length. Rng detects such cases where calling the MUA failed and puts the output in a file and asks the user kindly to attach it. Yes, it's rng's fault. I'm not sure why you say the shell has anything to do with this. How do you think rng invokes the different MUAs? If you have a better solution which supports calling MUAs with bodies of arbitrary textlengths, then please send me a patch. I'd love to have this included rather to rely on the user to attach a file. There are lots of ways to invoke a program and give it some data that don't involve giving that data to the shell. Cheers, Julien -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#422085: closed by Bastian Venthur [EMAIL PROTECTED] (Re: Bug#422085: bugscripts are executed, but content not put in email)
Julien Cristau wrote: On Mon, Jul 21, 2008 at 17:15:10 +0200, Bastian Venthur wrote: Julien Cristau wrote: That's one of the documented functions such a script can rely on. Where is that documented? And where does getkey come from? Haven't found a package providing this command. /usr/share/doc/reportbug/README.developers.gz The function is in /usr/share/reportbug/handle_bugscript which looks like it could be used pretty much as-is in rng. Probably, I'll see what I can do. as you can see rng already gathered the info and even put in in a file, all the user has to do is to attach this file to the mail. That doesn't make any sense. If you gather the info, then surely putting it in the mail is not much more work. I'm pretty sure you can attach files to a mail, if nothing else. Not every MUA supports an attach file option. Then they're broken. Ok, but that doesn't help very much, right? The file too long problem isn't rng's fault but a limitation of the shell which can't execute commands of unlimited length. Rng detects such cases where calling the MUA failed and puts the output in a file and asks the user kindly to attach it. Yes, it's rng's fault. I'm not sure why you say the shell has anything to do with this. How do you think rng invokes the different MUAs? If you have a better solution which supports calling MUAs with bodies of arbitrary textlengths, then please send me a patch. I'd love to have this included rather to rely on the user to attach a file. There are lots of ways to invoke a program and give it some data that don't involve giving that data to the shell. I'm still waiting for for a solution which works for python. Currently I use the command module of python which in turn calls os.popen. The limit of the command length seems to be somewhere between 1 and 2 characters (or maybe some completely different number) -- if you have a way to call a program with arbitrary command length, please tell me how -- I'll just apply the patch and we can forget about this whole please attach the file yourself thing. -- Bastian Venthur http://venthur.de Debian Developer venthur at debian org -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#422085: closed by Bastian Venthur [EMAIL PROTECTED] (Re: Bug#422085: bugscripts are executed, but content not put in email)
This one time, at band camp, Bastian Venthur said: And please keep in mind that this only happens when the output of a script is *very* large -- to large for the shell to handle -- like the one from xserver-xorg, for the most other packages, everything should work fine automatically and attaching a file is not needed. What does 'too large for the shell to handle' mean? Believe me, I've worked with the shell for quite a while now, and I have never seen a 'file too large' type error message since the bad old days when some filesystems didn't have LFS support. -- - | ,''`.Stephen Gran | | : :' :[EMAIL PROTECTED] | | `. `'Debian user, admin, and developer | |`- http://www.debian.org | - signature.asc Description: Digital signature
Bug#422085: closed by Bastian Venthur [EMAIL PROTECTED] (Re: Bug#422085: bugscripts are executed, but content not put in email)
On Mon, Jul 21, 2008 at 18:00:12 +0200, Bastian Venthur wrote: The output of /usr/share/bug/xserver-xorg/script 31 is ~65K big. Calling the mua like: mua --body $OUTPUT ... does not work, at least Calling the mua like that doesn't make any sense. Just give it a filename or something on stdin, don't inject arbitrary data in the command line. Puzzled, Julien -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#422085: closed by Bastian Venthur [EMAIL PROTECTED] (Re: Bug#422085: bugscripts are executed, but content not put in email)
Stephen Gran wrote: This one time, at band camp, Bastian Venthur said: And please keep in mind that this only happens when the output of a script is *very* large -- to large for the shell to handle -- like the one from xserver-xorg, for the most other packages, everything should work fine automatically and attaching a file is not needed. What does 'too large for the shell to handle' mean? Believe me, I've worked with the shell for quite a while now, and I have never seen a 'file too large' type error message since the bad old days when some filesystems didn't have LFS support. The output of /usr/share/bug/xserver-xorg/script 31 is ~65K big. Calling the mua like: mua --body $OUTPUT ... does not work, at least not with python's command module. Trying it directly in my shell results in the output to be truncated at a certain point, so I guess there is a limit. -- Bastian Venthur http://venthur.de Debian Developer venthur at debian org -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#422085: closed by Bastian Venthur [EMAIL PROTECTED] (Re: Bug#422085: bugscripts are executed, but content not put in email)
Julien Cristau wrote: On Mon, Jul 21, 2008 at 18:00:12 +0200, Bastian Venthur wrote: The output of /usr/share/bug/xserver-xorg/script 31 is ~65K big. Calling the mua like: mua --body $OUTPUT ... does not work, at least Calling the mua like that doesn't make any sense. Just give it a filename or something on stdin, don't inject arbitrary data in the command line. Please give me a concrete example, so I can implement your idea. Please test if it works with xdg-email and icedove. It also has to work combined with python's command.getoutput or at least from within Python. -- Bastian Venthur http://venthur.de Debian Developer venthur at debian org -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#422085: closed by Bastian Venthur [EMAIL PROTECTED] (Re: Bug#422085: bugscripts are executed, but content not put in email)
On Mon, Jul 21, 2008 at 18:16:39 +0200, Bastian Venthur wrote: Please give me a concrete example, so I can implement your idea. Please test if it works with xdg-email and icedove. It also has to work combined with python's command.getoutput or at least from within Python. *sigh* command=xdg-email ... if [ -d /usr/share/bug/$pkg ]; then script=/usr/share/bug/$pkg/script else script=/usr/share/bug/$pkg fi if [ -x $script ]; then foo=$(mktemp) /usr/share/reportbug/handle_bugscript $script $foo command=$command --attach $foo fi $command I don't know python, but I'm sure something like that is pretty easy. And no, I'm not going to test with icedove. Cheers, Julien -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#422085: closed by Bastian Venthur [EMAIL PROTECTED] (Re: Bug#422085: bugscripts are executed, but content not put in email)
Julien Cristau wrote: On Mon, Jul 21, 2008 at 18:16:39 +0200, Bastian Venthur wrote: Please give me a concrete example, so I can implement your idea. Please test if it works with xdg-email and icedove. It also has to work combined with python's command.getoutput or at least from within Python. *sigh* command=xdg-email ... if [ -d /usr/share/bug/$pkg ]; then script=/usr/share/bug/$pkg/script else script=/usr/share/bug/$pkg fi if [ -x $script ]; then foo=$(mktemp) /usr/share/reportbug/handle_bugscript $script $foo command=$command --attach $foo fi $command *sigh* Hmm ok, now we're starting to repeat ourself: - A few mails ago, I already explained (and provided a bugnumber), that the --attach parameter of xdg-email doesn't work and that not every MUA supports attaching a file from the command line. - You said, then the MUAs are broken. - I said, that conclusion doesn't help to solve the problem. The problem remains that I somehow have to get this info into the mail. I believe my current approach works good enough, but I'm of course open to better solutions. -- Bastian Venthur http://venthur.de Debian Developer venthur at debian org -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#422085: closed by Bastian Venthur [EMAIL PROTECTED] (Re: Bug#422085: bugscripts are executed, but content not put in email)
On Mon, Jul 21, 2008 at 18:46:16 +0200, Bastian Venthur wrote: - A few mails ago, I already explained (and provided a bugnumber), that the --attach parameter of xdg-email doesn't work and that not every MUA supports attaching a file from the command line. - You said, then the MUAs are broken. - I said, that conclusion doesn't help to solve the problem. I think it does. rng can choose not to use broken MUAs (which ones are these, btw?). Cheers, Julien -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#422085: closed by Bastian Venthur [EMAIL PROTECTED] (Re: Bug#422085: bugscripts are executed, but content not put in email)
Julien Cristau wrote: On Mon, Jul 21, 2008 at 18:46:16 +0200, Bastian Venthur wrote: - A few mails ago, I already explained (and provided a bugnumber), that the --attach parameter of xdg-email doesn't work and that not every MUA supports attaching a file from the command line. - You said, then the MUAs are broken. - I said, that conclusion doesn't help to solve the problem. I think it does. rng can choose not to use broken MUAs (which ones are these, btw?). That's not an option. I'd rather support more than less MUAs. Every MUA I'm aware of supports at least the mailto: paramter which is sufficient for rng's needs. The attachment option is *only* needed when the output of the script gets too long (so the mailto:-link gets too long, so the whole command gets too long for the shell) and therefore an attachment would be more suitable. There is no way I remove support for certain MUAs just because of this. The most imporant MUA which doesn't seem to work is xdg-email in combination with icedove set as default MUA (haven't tested others, but I suppose there are lots of others). -- Bastian Venthur http://venthur.de Debian Developer venthur at debian org -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
