Apportexpanded

Differences between revisions 1 and 25 (spanning 24 versions)
⇤ ← Revision 1 as of 2007-04-15 15:40:42 →
Size: 5629
Editor: 59
Comment:
← Revision 25 as of 2008-08-06 16:19:25 → ⇥
Size: 5255
Editor: localhost
Comment: converted to 1.6 markup
Deletions are marked like this. Additions are marked like this.
Line 1: Line 1:
 * '''Launchpad Entry''': UbuntuSpec:apport ftp-resumability-privacy
 * '''Created''': [[Date(2005-10-25T15:45:54Z)]]
 * '''Contributors''':
 * '''Packages affected''':
 * '''See also''': SpecTemplate		 ## page was renamed from Spec
 * '''Created''': <<Date(2007-04-15T21:40:54Z)>>
 * '''Contributors''':shirish
 * '''Launchpad entry''': [[https://bugs.launchpad.net/ubuntu/+source/apport/+bug/102868|Bug 102868]]
 * '''Packages affected''': [[https://bugs.launchpad.net/ubuntu/+source/apport|apport]]
 
Line 8: Line 9:
Add privacy, resumability and reliability features (with an ETA & upload speeds) to apport. More so for people with limited bandwidth ability and high latencies to server.
This will allow apport to be used more effectively. One can bank on apport to report all crashes without the user being interfered or being paranoid.
Line 9: Line 12:
This specification describes the way we would like Ubuntu specifications to be written. It takes the form of a specification itself.
Line 12: Line 14:
As of right now, one is never sure if apport has uploaded the bug or not. There are concerns regarding security, for e.g. if Firefox crashed or some application which has properties such as names, passwords, (in case of Firefox or other browsers) websites gone (profiling) etc. This needs to be communicated to the user. In essence apport has to be intuitive. Also people who have limited bandwidth capacity & have high latencies to the UK server are left stranded. An alternate to that would be setting up mirrors for crash reporting too.
Line 13: Line 16:
As we develop new ideas for features in Ubuntu, it's important to be able to communicate them clearly. This serves the purpose of making it clear what the feature is about, and allowing people to evolve an implementation strategy for it.

Publishing this content gives our community a chance to participate in the discussion and design of a feature, and increases the chance that community members will feel confident enough to start work on the implementation of the feature.

A good specification also allows community members who were not physically present at meetings discussing a topic to participate in the implementation of the spec.

Bottom line: the better your spec, the better the chances that your ideas will be clearly understood by the review team.
Line 23: Line 19:
  * Bob is the maintainer for the boot process for Ubuntu. In the Dapper cycle, he would like to work on getting the boot time down to two seconds from boot manager to GDM screen. He creates an entry for the specification in Launchpad, proposes it for the UBZ sprint, and starts writing out a braindump of it in the Ubuntu wiki. Magnus, who is in charge of UBZ scheduling, thinks it sounds fishy but approves it to make sure that the change is discussed and documented properly. He marks it as priority Medium because he isn't sure Bob will have time free for implementing it during Dapper.   * Bob application crashed. He has no idea how much time it would require for the uploading of crash to complete. Hence decisions (if his work is interrupted or finished) such as shutting down the machine, or going to hibernate or keep the machine running for the upload to finish for some time are unknown.
Line 25: Line 21:
  * Pedro works on Malone, in Launchpad. Before UBZ, he remembers that the dependency handling in the bug tracker is really not optimal. He writes out a Summary and Rationale in a Launchpad wiki page, registers it as a specification in Launchpad, and suggests it for UBZ. Monica, Launchpad manageress, thinks that this is really not the time to be talking about it and rejects the application for UBZ. He then indicates it for the next conference, UBB, and marks its priority is Low.   * Peter application crashes. He thinks apport would report it. During the upload for some reason or the other, network connection lost, power cut-out etc, he is unable to upload the crash file. The crash file remains in /var/crash & is not useful for anybody.
Line 27: Line 23:
  * Jason is an Ubuntu and Rosetta user. He has noticed that changes made to translations are making their way into language packs but not to the upstream versions, and adds a specification that describes a way for getting upstream to use language packs. Monica also has a plan for this but hadn't described it in a spec, so she adds it to the UBZ spec list, and adds Carlos, Rosetta maintainer, as drafter for it.   * Jason is working in a bank/clinic/finance instituion. He was making an e-commerce transaction through some application (for e.g. Firefox) and the application crashed. He is in 2 minds whether to send the crash or not.

  * shirish tries to upload a crash file. After waiting for about an hr. He gets a message saying HTTP Error 500: Internal Server Error. [[https://bugs.launchpad.net/ubuntu/+source/apport/+bug/99989]]
Line 31: Line 30:
This specification covers feature specifications for Ubuntu and Launchpad. It is not meant as a more general specification format. Any user using apport Ubuntu Edgy, Feisty & above.
Line 35: Line 34:
A specification should be built with the following considerations: === Big Picture ===
 This will allow apport to be a more robust application.
Line 37: Line 37:
  * The person implementing it may not be the person writing it. It should be clear enough for someone to be able to read it and have a clear path towards implementing it. If it doesn't, it needs more detail.  *Uploading features
  * This will allow apport to show an an upload bar with an ETC for bugs. Similar to how its done at rapidshare.com or any other public upload site. Also have some kind of checksum algorithm built-in at the time of sending. This should be either based on user's bandwidth or set by him. (please clarify)
Line 39: Line 40:
  * That the use cases covered in the specification should be practical situations, not contrived issues.  *Resumability
  * This will make sure /var/crash remains empty at all times. If possible send crashes when network activity is at the lower/lowest. The idea is of actually resuming (and not restarting) previous uploads: that is, continuing where the user left off, instead of starting another hour-plus upload, finishing the final five minutes that the server is missing & completing the bug-report & giving user the chance of adding more info. Less un-owned crash reports where there are no details etc.
Line 41: Line 43:
  * Limitations and issues discovered during the creation of a specification should be clearly pointed out so that they can be dealt with explicitly.  *Messages
  *Give user error messages otherwise remain in background. If the user wants a verbose mode then show him all the goodies. Alternatively send messges to root user via mail or some other place.
Line 43: Line 46:
  * If you don't know enough to be able to competently write a spec, you should either get help or research the problem further. Avoid spending time making up a solution: base yourself on your peers' opinions and prior work.

Specific issues related to particular sections are described further below.

=== Summary ===

The summary should not attempt to say '''why''' the spec is being defined, just '''what''' is being specified.

=== Rationale ===

This should be the description of '''why''' this spec is being defined.

=== Scope and Use Cases ===

While not always required, but in many cases they bring much better clarity to the scope and scale of the specification than could be obtained by talking in abstract terms.		  *Logs
  *The apport log should also have error messages as to why a certain report could not be sent. If user has turned off apport it should be mentioned.
Line 61: Line 51:
This section is usually broken down into subsections, such as the packages being affected, data and system migration where necessary, user interface requirements and pictures (photographs of drawings on paper work well).  Packages effected :-

  *apport-gtk
  *apport-qt and probably
  *python-apport
  *python-problem-report
    
Line 65: Line 62:
To implement a specification, the assignee should observe the use cases carefully, and follow the design specified. He should make note of places in which he has strayed from the design section, adding rationale describing why this happened. This is important so that next iterations of this specification (and new specifications that touch upon this subject) can use the specification as a reference.  Atleast on the upload progress bar, somebody has done something similar. http://www.opensky.ca/~jdhildeb/software/kodakloader/
The last 2 paragraphs are esp. interesting :-
Line 67: Line 65:
The implementation is very dependent on the type of feature to be implemented. Refer to the team leader for further suggestions and guidance on this topic. I used John J. Lee's mechanize, ClientForm and ClientCookie modules to emulate a browser. Hopefully this won't break too often if they rename URLs and links (KodakLoader does some screen-scraping to read out existing photo albums). I'd expect that Kodak Gallery has a soap or xmlrpc interface (which their ofoto software uses), so it'd be great if they were nice enough to provide some documentation about this (I haven't asked, but am considering doing so).
Probably the single largest hurdle was figuring out how to report accurate progress for the file upload, as the actual upload is buried within mechanize, urllib and httplib. Once I was able to figure out the twisty maze, it didn't take too much code to implement. I used some hooks to provide a few alternate classes, which enabled me to break up the transfer into chunks and invoke a callback occasionally.
Line 71: Line 71:
The specification process requires experienced people to drive it. More documentation on the process should be produced.
Line 73: Line 72:
The drafting of a specification requires English skills and a very good understanding of the problem. It must also describe things to an extent that someone else could implement. This is a difficult set of conditions to ensure throughout all the specifications added.
Line 75: Line 73:
There is a lot of difficulty in gardening obsolete, unwanted and abandoned specifications in the Wiki.
Line 79: Line 78:
We'll have a first public session on this on the first Monday in UBZ. == References ==

 * [[http://www.rapidshare.com|Rapidshare]]
 * [[http://images.rapidshare.com/software/rapiduploader/RapidUploader.exe|Rapiduploader]]
 * [[http://en.wikipedia.org/wiki/Comparison_of_one-click_hosters|One click hosters having upload tools]]

 

== Comments ==
Line 81: Line 90:
CategorySpec CategorySpec CategoryBugSquad

  • Created: 2007-04-15

  • Contributors:shirish

  • Launchpad entry: Bug 102868

  • Packages affected: apport

Summary

Add privacy, resumability and reliability features (with an ETA & upload speeds) to apport. More so for people with limited bandwidth ability and high latencies to server. This will allow apport to be used more effectively. One can bank on apport to report all crashes without the user being interfered or being paranoid.

Rationale

As of right now, one is never sure if apport has uploaded the bug or not. There are concerns regarding security, for e.g. if Firefox crashed or some application which has properties such as names, passwords, (in case of Firefox or other browsers) websites gone (profiling) etc. This needs to be communicated to the user. In essence apport has to be intuitive. Also people who have limited bandwidth capacity & have high latencies to the UK server are left stranded. An alternate to that would be setting up mirrors for crash reporting too.

Use Cases

  • Bob application crashed. He has no idea how much time it would require for the uploading of crash to complete. Hence decisions (if his work is interrupted or finished) such as shutting down the machine, or going to hibernate or keep the machine running for the upload to finish for some time are unknown.

  • Peter application crashes. He thinks apport would report it. During the upload for some reason or the other, network connection lost, power cut-out etc, he is unable to upload the crash file. The crash file remains in /var/crash & is not useful for anybody.

  • Jason is working in a bank/clinic/finance instituion. He was making an e-commerce transaction through some application (for e.g. Firefox) and the application crashed. He is in 2 minds whether to send the crash or not.

  • shirish tries to upload a crash file. After waiting for about an hr. He gets a message saying HTTP Error 500: Internal Server Error. https://bugs.launchpad.net/ubuntu/+source/apport/+bug/99989

Scope

Any user using apport Ubuntu Edgy, Feisty & above.

Design

Big Picture

  • This will allow apport to be a more robust application.
  • Uploading features
    • This will allow apport to show an an upload bar with an ETC for bugs. Similar to how its done at rapidshare.com or any other public upload site. Also have some kind of checksum algorithm built-in at the time of sending. This should be either based on user's bandwidth or set by him. (please clarify)
  • Resumability

    • This will make sure /var/crash remains empty at all times. If possible send crashes when network activity is at the lower/lowest. The idea is of actually resuming (and not restarting) previous uploads: that is, continuing where the user left off, instead of starting another hour-plus upload, finishing the final five minutes that the server is missing & completing the bug-report & giving user the chance of adding more info. Less un-owned crash reports where there are no details etc.

  • Messages
    • Give user error messages otherwise remain in background. If the user wants a verbose mode then show him all the goodies. Alternatively send messges to root user via mail or some other place.
  • Logs
    • The apport log should also have error messages as to why a certain report could not be sent. If user has turned off apport it should be mentioned.

Implementation Plan

  • Packages effected :-
    • apport-gtk
    • apport-qt and probably
    • python-apport
    • python-problem-report

Implementation

The last 2 paragraphs are esp. interesting :-

I used John J. Lee's mechanize, ClientForm and ClientCookie modules to emulate a browser. Hopefully this won't break too often if they rename URLs and links (KodakLoader does some screen-scraping to read out existing photo albums). I'd expect that Kodak Gallery has a soap or xmlrpc interface (which their ofoto software uses), so it'd be great if they were nice enough to provide some documentation about this (I haven't asked, but am considering doing so). Probably the single largest hurdle was figuring out how to report accurate progress for the file upload, as the actual upload is buried within mechanize, urllib and httplib. Once I was able to figure out the twisty maze, it didn't take too much code to implement. I used some hooks to provide a few alternate classes, which enabled me to break up the transfer into chunks and invoke a callback occasionally.

Outstanding Issues

BoF agenda and discussion

References

Comments

CategorySpec CategoryBugSquad

Apportexpanded (last edited 2008-08-06 16:19:25 by localhost)