wow, really sourceforge?

I filed a bug on Notepad++ today.

The first attempt silently failed – it just took me back to the ‘browse’ screen with no error message (that I could find). Part of what I had included was a PSR output zip file (a little under 1MB – I love it when people that file bugs I handle use that or something similar :), which at the moment seems to have been what caused the bug not to create the first time (although that would have been nice to have known before losing all the data I had put in) Thankfully I’ve gotten bitten by similar experiences with other sites so I had copied the description text out first. Smile

On the second attempt, I didn’t try to include the file upload, I just put in the various text values.  That created fine (the bug is here if you want to read it).  Then I tried an ‘update’ to attach the PSR file:

image

256KBReally?

I also had created a dump file (Notepad++ hangs in the repro) which is 96MB – I expected that it might be too large for SourceForge to accept, but in 2011 I didn’t expect to find any sites that accepted user uploads and limited it to 256KB.  Disks are cheap, S3 is cheap, etc. – spam me with some ads (ok, MORE ads – SF’s already pretty spammy as-is) if you need to try and bring in a few extra bucks to cover the cost or something.  Sheesh.

Now I have to figure out what other method I want to use (dropbox? livedrive? any recommendations?) to share these files and I have to track the bug to know when it’s been fixed and I can safely remove the files.

Caveat: Now, in this particular case, I think (hope?) the repro steps I wrote are clear enough and should be easy enough to repro that I don’t think the PSR output or dump file will actually be required to debug/fix this issue, but certainly any project hosting site that makes it painful for those trying to file good bugs to do so is doing the projects it hosts (and the communities around those projects) a great disservice.

Advertisements

String or binary data would be truncated.

I can’t tell you how much time and effort (not to mention frustration) would be saved (and would have been saved) if SQL Server would include the table+column in the error.  It’s very frustrating to have a transaction fail (especially when it’s one in the middle of a long run of transactions, when all the others have worked fine).

DBA’s and developers have complained about this for over a decade, but no such luck in getting it actually fixed, at least as of 2008 R2.

I’m not holding out much hope for Denali changing the situation, either Sad smile

Dear Reads: Any tips/tricks to get the offending column insert/update? 

As-is I run the sql profiler, find the transaction that fails, then with a copy of the database I try the individual statements and commit each until I hit the issue (and hopefully there’s only one issue to fix, but if not, we repeat the whole thing).

NOTE: yes, I know that you can turn off ANSI_WARNINGS and SQL Server will do the truncate for you, but I’d much rather find out the offending column and update the code to handle it (for instance, we might do an ellipsis-based approach).

There’s gotta be a better way ™ Smile