The problem with high CPU utilization is that the MergePDF process is being run from our production content server. All content is fed live from the server, and high utlitization can really slow down delivery of form kits. Then users refresh and start the process over again, only compounding the problem. If it takes 10 seconds to build the kit, it is too long.

But on the other hand, any CPU cycle not used is forever lost. Same goes for bandwidth.

Ken,

Good work. Useful information.

In software development there are two things I rarely worry about.

1.)     Disk space. By the time you are close to filling up you 200GB drive the 2TB drives will $98.99 plus tax at CompUSA

2.)     CPU Usage. Same reason as the drives.

It’s probably a bad practice however I don’t spend a lot of time optimizing code when hardware will have solved most of my problems by time I’m ready to release.

Here is what has been discovered with lots of testing:

There are basically 4 ways to merge PDF files and they yield different results.  MergeDocument() in memory, mergeFiles(), mergeFileList(), and mergeFileListFast().

MergeDocument() in memory is the fast and correctly appends redundant field objects correctly.

MergeFiles() and MergeFileList() are very slow, and have mixed reults with redundant fields.

MergeFilesFast() is fast, but does not handle redundant fields well at all.

After any of these calls, if you issue a SetNeedsAppearance() call, it will trigger acrobat to "fix up form fields" on load of the PDF, allowing redundant field objects to properly resolve. However, this is not a solution if you want the program to populate field data before delivery because the "fix up fields" is run after the file is delivered to the viewer.

The solution I have come up with is using the mergeDocument() method, which is fairly fast but very CPU intensive. But it leaves the PDF in a state in which you can properly pump data into it.

Right now this is working, but I will be closely monitoring the CPU usage for heavy loads. Thanks to all who contributed, expecially the contributor who mentioned the SetNeedsAppearance() method in another post.

Here is an update:

When I have multiple PDFs with same named field on them, and they are merged, the field on pages following the 1st occurance is disabled and causes acrobat5 to crash.

We have discovered if the PDF is opened in Acrobat, remove security, and Extract All the pages, the subsequent form fields are then active and fillable.

The Extract PAges seems to "reform" the PDF file. The status bad indicates "Fixing Up Form Field" during the process.

Does anyone know what this is fixing, and is there an equivalent QP process?  I have tried the CheckObjects() and ExtractPAges(), but the do not seem to have the desired effect.

Send Me a couple of files and a quick example of the code you use to perform the merge. I will watch the process in the debugger. I may not be able to tell how damage is occurring but I should be able to tell you what part of the process take the most time.

Steve

I need to get a better understanding of how PDF files are merged. We have been having terrible problems with form kits that are generated by QuickPDF since we started adding lots of fields.

We have had a process in place for years that combined multiple PDFs into a single download pdf. It had worked very well using the merging in memory.

Since then we have added fillable fields to many of the PDF files, some with a couple hundred fields. The process kept getting slower as files sizes increased.  I rewrote the process recently to merge using the MergeFileListFast() method, but it is still not as fast as we expected.

The biggest problem is that if we have different forms with fields that are named the same, once they are merged, the second occurance of the field is no longer fillable. These merged files actually crash Acrobat 5.

I need a better understanding of exactly what is done during a merge. Is is simple contatenation of the files? Or are they really merged into a single PDF "container".  Is there a process that can be done to better combine the files so fields work as though they are in a single document?

I can provide sample PDF files if someone has a way to analyze them and tell me what is damaged by the process.