Issue with the autobuilders: removing 1379 builds from the history?
Hello,
While trying to reproduce the flex recent issue that triggered in the autobuilders, I found out that I couldn't reproduce it. After investigating, I discovered that the Git commit id being reported by my autobuilder machine was incorrect. This is due to a change I made in the autobuilder on May, 13th, where I'm now using a separate Git tree for each build instead of a common one, but the script was still taking the Git commit id of some other Buildroot tree. So the build was actually taking place with the latest Buildroot version, but reported to use Git commit id 9947ee9fbb94af800b8b054262734e0f81f32250.
This means that when you try to reproduce the build, if you take the Git commit id given by the autobuilder website, then you won't be testing with the Git commit id with which the build was made, which is quite annoying. This is what happened to me when trying to reproduce the flex problem: I have a script that, given a build SHA1, downloads the configuration and Git commit ID, and restarts the build with the exact same commit and the exact same configuration. But for those builds who reported an invalid Git commit ID... it cannot work, and therefore I wasn't reproducing the flex problem.
Therefore, I am considering removing those completely bogus build results from the history. We are talking about removing 1379 build results, i.e all builds done my autobuilder between May, 14th and today.
Even though it is quite annoying to lose so many results, I believe this is the best action we can take now.
What do you think?
Thomas
Signature · 4 lines
Thomas Petazzoni, Free Electrons Kernel, drivers, real-time and embedded Linux development, consulting, training and support. http://free-electrons.com
On 2013-05-28 09:10 +0200, Thomas Petazzoni spake thusly:
Quoted 9 lines
While trying to reproduce the flex recent issue that triggered in the autobuilders, I found out that I couldn't reproduce it. After investigating, I discovered that the Git commit id being reported by my autobuilder machine was incorrect. This is due to a change I made in the autobuilder on May, 13th, where I'm now using a separate Git tree for each build instead of a common one, but the script was still taking the Git commit id of some other Buildroot tree. So the build was actually taking place with the latest Buildroot version, but reported to use Git commit id
9947ee9fbb94af800b8b054262734e0f81f32250.
[--SNIP--]
Quoted 6 lines
Therefore, I am considering removing those completely bogus build results from the history. We are talking about removing 1379 build results, i.e all builds done my autobuilder between May, 14th and today.
Even though it is quite annoying to lose so many results, I believe this is the best action we can take now.
No point in keeping those since we can't reproduce them, I'd say.
Regards,
Yann E. MORIN.
Signature · 6 lines
.-----------------.--------------------.------------------.--------------------. | Yann E. MORIN | Real-Time Embedded | /"\ ASCII RIBBON | Erics' conspiracy: | | +33 662 376 056 | Software Designer | \ / CAMPAIGN | ___ | | +33 223 225 172 `------------.-------: X AGAINST | \e/ There is no | | http://ymorin.is-a-geek.org/ | _/*\_ | / \ HTML MAIL | v conspiracy. | '------------------------------^-------^------------------^--------------------'
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
"Yann" == Yann E MORIN <[email protected]> writes:
Hi,
Quoted 6 lines
Quoted 6 lines
Therefore, I am considering removing those completely bogus build results from the history. We are talking about removing 1379 build results, i.e all builds done my autobuilder between May, 14th and today.
Even though it is quite annoying to lose so many results, I believe this is the best action we can take now.
Yann> No point in keeping those since we can't reproduce them, I'd say.
Agreed.
Signature · 1 line
Bye, Peter Korsgaard
Quoted 10 lines · Thomas Petazzoni
This means that when you try to reproduce the build, if you take the Git commit id given by the autobuilder website, then you won't be testing with the Git commit id with which the build was made, which is quite annoying. This is what happened to me when trying to reproduce the flex problem: I have a script that, given a build SHA1, downloads the configuration and Git commit ID, and restarts the build with the exact same commit and the exact same configuration. But for those builds who reported an invalid Git commit ID... it cannot work, and therefore I wasn't reproducing the flex problem.
Quoted 6 lines
Therefore, I am considering removing those completely bogus build results from the history. We are talking about removing 1379 build results, i.e all builds done my autobuilder between May, 14th and today.
Even though it is quite annoying to lose so many results, I believe this is the best action we can take now.
I would wait a couple of days before removing them. At least for the
most recent ones, current HEAD probably still exposes the same problem, so in that sense the build result is still usable.
Would it be possible to direct the URLs to a placeholder page
explaining that the build result was removed, so existing links in commit messages don't die?
Regards,
Arnout
Signature · 6 lines
Arnout Vandecappelle arnout at mind be Senior Embedded Software Architect +32-16-286500 Essensium/Mind http://www.mind.be G.Geenslaan 9, 3001 Leuven, Belgium BE 872 984 063 RPR Leuven LinkedIn profile: http://www.linkedin.com/in/arnoutvandecappelle GPG fingerprint: 7CB5 E4CC 6C2E EFD4 6E3D A754 F963 ECAB 2450 2F1F
Dear Arnout Vandecappelle,
Quoted 3 lines · Arnout Vandecappelle
I would wait a couple of days before removing them. At least for the
most recent ones, current HEAD probably still exposes the same problem, so in that sense the build result is still usable.
Ok. I'm not sure they are really usable, but ok.
Quoted 3 lines
Would it be possible to direct the URLs to a placeholder page
explaining that the build result was removed, so existing links in commit messages don't die?
Yes, I can keep the output directories around, wipe-out their contents, and replace it with just a README.txt file that says it has been removed and why. Would this be ok for you?
Best regards,
Thomas
Signature · 4 lines
Thomas Petazzoni, Free Electrons Kernel, drivers, real-time and embedded Linux development, consulting, training and support. http://free-electrons.com
Quoted 9 lines · Thomas Petazzoni
Dear Arnout Vandecappelle,
Quoted 3 lines · Arnout Vandecappelle
I would wait a couple of days before removing them. At least for the
most recent ones, current HEAD probably still exposes the same problem, so in that sense the build result is still usable.Ok. I'm not sure they are really usable, but ok.
I usually don't look at the commit ID when reproducing an autobuilder
failure. If I haven't seen a related patch on the list, I'm pretty sure it will still fail.
Quoted 7 lines
Quoted 3 lines
Would it be possible to direct the URLs to a placeholder page
explaining that the build result was removed, so existing links in commit messages don't die?Yes, I can keep the output directories around, wipe-out their contents, and replace it with just a README.txt file that says it has been removed and why. Would this be ok for you?
Perfect!
Regards,
Arnout
Signature · 6 lines
Arnout Vandecappelle arnout at mind be Senior Embedded Software Architect +32-16-286500 Essensium/Mind http://www.mind.be G.Geenslaan 9, 3001 Leuven, Belgium BE 872 984 063 RPR Leuven LinkedIn profile: http://www.linkedin.com/in/arnoutvandecappelle GPG fingerprint: 7CB5 E4CC 6C2E EFD4 6E3D A754 F963 ECAB 2450 2F1F