Closed adamwitwer closed 12 years ago
Here's the error:
Traceback (most recent call last):
File "/usr/local/app/worker/atlas-workers/workers/err.py", line 19, in <module>
print "%s\n" % linecache.getline(xml_source,int(fields[1]))
ValueError: invalid literal for int() with base 10: 'rated for 0.1μF at 50V. Center'
Still waiting for files to download....
Thanks Adam. Adam/Sanders do you know what that error is?
On Mon, Sep 24, 2012 at 10:09 AM, Adam Witwer notifications@github.comwrote:
Here's the error:
Traceback (most recent call last): File "/usr/local/app/worker/atlas-workers/workers/err.py", line 19, in
print "%s\n" % linecache.getline(xml_source,int(fields[1])) ValueError: invalid literal for int() with base 10: 'rated for 0.1μF at 50V. Center' Still waiting for files to download....
— Reply to this email directly or view it on GitHubhttps://github.com/MakerPress/atlas-public-feedback/issues/88#issuecomment-8819884.
Thanks, Adam. Looks like this is some weird handling of the literal in the code that parses the asciidoc error log. Looking into is...
Andrew
ps -- weird that it's taking you so long to download. Is it done? We've noticed that the new atlas-admin seems to slap requests into the "denyhosts" bucket pretty quickly. If it's not done soon let me know and I'll check it out. It could just be that it's a huge file that's taking a while...
----- Original Message ----- From: Adam Witwer notifications@github.com To: MakerPress/atlas-public-feedback atlas-public-feedback@noreply.github.com Sent: Mon, 24 Sep 2012 10:09:30 -0400 (EDT) Subject: Re: [atlas-public-feedback] Build error (#88)
Here's the error:
Traceback (most recent call last):
File "/usr/local/app/worker/atlas-workers/workers/err.py", line 19, in <module>
print "%s\n" % linecache.getline(xml_source,int(fields[1]))
ValueError: invalid literal for int() with base 10: 'rated for 0.1μF at 50V. Center'
Still waiting for files to download....
Reply to this email directly or view it on GitHub:
Andrew Odewahn Chief Technology Officer O'Reilly Media 617-460-6693
Still going! It's our old friend, the Encyclopedia of Electronic Components. Only 43% of the way there....
$ git clone https://adam%40oreilly.com@atlas-server.oreilly.com/git/1230000000669.git encyc_elec_comp Initialized empty Git repository in /mnt/work/work/tools/adam/books/encyc_elec_comp/.git/ Password: remote: Counting objects: 4133, done. remote: Compressing objects: 100% (4000/4000), done. Receiving objects: 43% (1803/4133), 1.27 GiB | 678 KiB/s
On Sep 24, 2012, at 10:20 AM, Andrew Odewahn wrote:
Thanks, Adam. Looks like this is some weird handling of the literal in the code that parses the asciidoc error log. Looking into is...
Andrew
ps -- weird that it's taking you so long to download. Is it done? We've noticed that the new atlas-admin seems to slap requests into the "denyhosts" bucket pretty quickly. If it's not done soon let me know and I'll check it out. It could just be that it's a huge file that's taking a while...
----- Original Message ----- From: Adam Witwer notifications@github.com To: MakerPress/atlas-public-feedback atlas-public-feedback@noreply.github.com Sent: Mon, 24 Sep 2012 10:09:30 -0400 (EDT) Subject: Re: [atlas-public-feedback] Build error (#88)
Here's the error:
Traceback (most recent call last): File "/usr/local/app/worker/atlas-workers/workers/err.py", line 19, in <module> print "%s\n" % linecache.getline(xml_source,int(fields[1])) ValueError: invalid literal for int() with base 10: 'rated for 0.1μF at 50V. Center'
Still waiting for files to download....
Reply to this email directly or view it on GitHub:
https://github.com/MakerPress/atlas-public-feedback/issues/88#issuecomment-8819884
Andrew Odewahn Chief Technology Officer O'Reilly Media 617-460-6693 — Reply to this email directly or view it on GitHub.
The plot thickens. It looks like there are some entities ( Ω and μ) being used in the xml that the xml parser doesn't like. Here are a few examples from the error logs of the worker:
/usr/local/app/worker/tmp/pdf/1230000000669/book.xml:2528: parser error : Entity 'Omega' not defined asis>. In these examples, each has an arbitrary rated resistance of 4,700Ω ^ /usr/local/app/worker/tmp/pdf/1230000000669/book.xml:2812: parser error : Entity 'mu' not defined ITOR_CERAMIC_GENERIC">
There's still a problem with the code that's parsing the error log, but this is the underlying problem it's trying to report.
----- Original Message ----- From: Adam Witwer notifications@github.com To: MakerPress/atlas-public-feedback atlas-public-feedback@noreply.github.com Cc: Andrew Odewahn odewahn@oreilly.com Sent: Mon, 24 Sep 2012 10:22:04 -0400 (EDT) Subject: Re: [atlas-public-feedback] Build error (#88)
Still going! It's our old friend, the Encyclopedia of Electronic Components. Only 43% of the way there....
$ git clone https://adam%40oreilly.com@atlas-server.oreilly.com/git/1230000000669.git encyc_elec_comp Initialized empty Git repository in /mnt/work/work/tools/adam/books/encyc_elec_comp/.git/ Password: remote: Counting objects: 4133, done. remote: Compressing objects: 100% (4000/4000), done. Receiving objects: 43% (1803/4133), 1.27 GiB | 678 KiB/s
On Sep 24, 2012, at 10:20 AM, Andrew Odewahn wrote:
Thanks, Adam. Looks like this is some weird handling of the literal in the code that parses the asciidoc error log. Looking into is...
Andrew
ps -- weird that it's taking you so long to download. Is it done? We've noticed that the new atlas-admin seems to slap requests into the "denyhosts" bucket pretty quickly. If it's not done soon let me know and I'll check it out. It could just be that it's a huge file that's taking a while...
----- Original Message ----- From: Adam Witwer notifications@github.com To: MakerPress/atlas-public-feedback atlas-public-feedback@noreply.github.com Sent: Mon, 24 Sep 2012 10:09:30 -0400 (EDT) Subject: Re: [atlas-public-feedback] Build error (#88)
Here's the error:
Traceback (most recent call last): File "/usr/local/app/worker/atlas-workers/workers/err.py", line 19, in <module> print "%s\n" % linecache.getline(xml_source,int(fields[1])) ValueError: invalid literal for int() with base 10: 'rated for 0.1μF at 50V. Center'
Still waiting for files to download....
Reply to this email directly or view it on GitHub:
https://github.com/MakerPress/atlas-public-feedback/issues/88#issuecomment-8819884
Andrew Odewahn Chief Technology Officer O'Reilly Media 617-460-6693 — Reply to this email directly or view it on GitHub.
Reply to this email directly or view it on GitHub:
Andrew Odewahn Chief Technology Officer O'Reilly Media 617-460-6693
Yes, I want to take a stab at making sure we always bubble these errors back to Atlas. That also means never to have an eternal "pending" job.
Man, that's a biiiiig book.
Okay, so the book-docinfo.xml (used for book metadata and copyright page stuff) was very, very badly formed. I fixed that. Builds still failing, but I'm getting closer. I think the entity errors were bogus, caused by the badly formed book-docinfo.xml file.
On Sep 24, 2012, at 10:41 AM, Rune Skjoldborg Madsen wrote:
Yes, I want to take a stab at making sure we always bubble these errors back to Atlas. That also means never to have an eternal "pending" job.
Man, that's a biiiiig book.
— Reply to this email directly or view it on GitHub.
Okay, PDFs are building again. In this case, there was no way for a user working strictly in Atlas to locate and fix the errors.
Also, I noticed that this book had some less severe validity errors. Because we're using the --invalid-ok option, the user is not aware of the validity errors, and a book could potentially make it's way through production as invalid XML, which will cause problems down the line. We should think about a adding build warnings to Atlas, which would list validity errors, even when a build doesn't fail.
On Sep 24, 2012, at 10:54 AM, Adam Witwer wrote:
Okay, so the book-docinfo.xml (used for book metadata and copyright page stuff) was very, very badly formed. I fixed that. Builds still failing, but I'm getting closer. I think the entity errors were bogus, caused by the badly formed book-docinfo.xml file.
On Sep 24, 2012, at 10:41 AM, Rune Skjoldborg Madsen wrote:
Yes, I want to take a stab at making sure we always bubble these errors back to Atlas. That also means never to have an eternal "pending" job.
Man, that's a biiiiig book.
— Reply to this email directly or view it on GitHub.
Melanie asked me about a build error:
http://atlas.labs.oreilly.com/books/168/jobs/10964
My suspicion is that there is a build error and that Atlas isn't passing the error back correctly. But that's just a guess. I'm cloning the files so that I can run pdfrenderer independently of Atlas to identify the error. It's taking FOREVER to clone because the repo is so large.