Reduce size of large commit messages about bfg-repo-cleaner HOT 3 OPEN

If the option accepts perl-compatible regex, then this is possible and simple:

--replace-message-text 'qr{^.{1000000,1000000000}$}s'

the above will match msg lines that are > 1Mb and < 1Gb in size.
Note that we found another instance where a single short msg line has been
repeated millions of times.

Thanks

Richard

On Mon, Jul 15, 2013 at 12:29:33AM -0700, Roberto Tyley wrote:

@chenri mentions in #20 that his repo contains a very large commit message:

due to someone repeating a short line a huge number of times (editing
error?) so that a single msg line was 6Mb long. We never detected this
until we tried bfg last week.

It would be nice if The BFG had the ability to somehow reduce the size of large
commit messages - but how exactly would it do it? Some options:

• Simply truncate the entire message after the first X KB?
• Allow user to run a --replace-message-text option, similar to
--replace-text?? This would work for a repeated value on a single line, so
long as the line wasn't too long, but how would it work on a commit message
with 1 million lines, each distinct?

—
Reply to this email directly or view it on GitHub.*

from bfg-repo-cleaner.

The BFG uses regex expressions provided by the Java SDK, which are quite extensive:

http://docs.oracle.com/javase/7/docs/api/java/util/regex/Pattern.html

I'll hopefully get some time in the next 24 hours to look at this.

from bfg-repo-cleaner.

Hi Roberto,

Did you get a chance to work on this? It would be really

nice if bfg-repo-cleaner can handle huge msg size.

Thanks

Richard

On Tue, Jul 16, 2013 at 02:09:46AM -0700, Roberto Tyley wrote:

The BFG uses regex expressions provided by the Java SDK, which are quite
extensive:

http://docs.oracle.com/javase/7/docs/api/java/util/regex/Pattern.html

I'll hopefully get some time in the next 24 hours to look at this.

—
Reply to this email directly or view it on GitHub.*

from bfg-repo-cleaner.