X-Git-Url: https://wimlib.net/git/?p=wimlib;a=blobdiff_plain;f=doc%2Fimagex-capture.1.in;h=a0927133c33f02d05d78d4fc94aa2e5c0d1bc33a;hp=c29cf6059e257ae44f773b97c8bf7257d2c8cef4;hb=88f5356383a47f77185c8c1723ce9eac7da017ee;hpb=213d64619675a59db71fed619bf47c1870b0e2e2

diff --git a/doc/imagex-capture.1.in b/doc/imagex-capture.1.in
index c29cf605..a0927133 100644
--- a/doc/imagex-capture.1.in
+++ b/doc/imagex-capture.1.in
@@ -1,4 +1,4 @@
-.TH IMAGEX "1" "October 2012" "imagex (wimlib) wimlib @VERSION@" "User Commands"
+.TH IMAGEX "1" "November 2012" "imagex (wimlib) wimlib @VERSION@" "User Commands"
 .SH NAME
 imagex-capture, imagex-append \- Capture a WIM image from a directory tree
 
@@ -119,6 +119,12 @@ processors).  Note: if creating or appending to an uncompressed WIM, additional
 threads will not be used, regardless of this parameter, since no compression
 needs to be done in this case.
 .TP
+\fB--rebuild\fR
+For \fBimagex append\fR: rebuild the entire WIM rather than appending the new
+data to the end of it.  Rebuilding the WIM is slower, but will save a little bit
+of space that would otherwise be left as a hole in the WIM.  Also see \fBimagex
+optimize\fR.
+.TP
 \fB--flags\fR=\fIEDITIONID\fR
 Specify a string to use in the <FLAGS> element of the XML data for the new
 image.
@@ -194,6 +200,24 @@ type, and \fBimagex\fR always enforces this.
 stream chunk size of 32768.  The only WIMs I've seen that are different from
 this are some pre-Vista WIMs that had a different version number.
 
+Unless \fB--rebuild\fR is specified, aborting an \fBimagex append\fR command
+mid-way through has a small chance of corrupting the WIM file.  However, a
+precaution is taken against this, so it should be very unlikely.  In the event
+of an aborted \fBimagex append\fR, \fBimagex optimize\fR may be run to remove
+extra data that may have been partially appended to the physical WIM file but
+not yet incorporated into the structure of the WIM.
+
+Capturing or appending an image happens in two main phases: (1) scanning the
+directory or NTFS volume to checksum all the files and determine the streams to
+be written, and (2) writing the new streams to the WIM file.  Streams are not
+stored in memory after (1), since there could easily be gigabytes of data;
+instead, they are read again during step (2); however, duplicate streams in the
+image, and streams already existing in any other image in the WIM, are not read
+again.  In the future, it may be possible to introduce the ability to capture an
+image with reading each file only one time, although this mode would have some
+limitations--- for example, a stream might be compressed only to be thrown away
+as a duplicate once it's been checksummed.
+
 .SH EXAMPLES
 Create a new WIM 'mywim.wim' from the directory 'somedir', using LZX compression and
 including an integrity table: