Ask HN: Is PDF generation a real pain for developers?

10 points by knkella 13 years ago | 30 comments

Recently we launched a PDF generation service on cloud by the name of Pagify http://pagify.io for developers. Our main assumptions were:

1) It is tedious to code the layout and design of the PDF using code libraries such as FPDF, Prawn etc.

2) When it came to generating large no of PDFs it is difficult to scale.

3) Developer wants to focus on the core offering of their web product rather that spending their time on stuff like PDF generation, email, image processing, payments etc.

4) In cloud development model off-the-shelf components are preferred in the form of cloud services or cloud API over code packages or software. So PDF generation offered as a service instead of code library is better.

Targeting these assumptions Pagify offers a designer within the browser where users can design their PDF template and use an API to generate PDFs.

But after launching on HN and betalist we received mixed feedback on Pagify. Do you think our assumptions are correct? And does Pagify solve this problem?

Thanks, Kush.

danielstudds 13 years ago |

I work in the DOCCM (Document Output for Customer Communications Management) space, which in a certain light is just generating PDFs (well, perhaps a bit more complex than that...). Enterprise deals in this space will have 6 or 7 zeros, so yes, there is a market for generating PDFs. There are already some cloud providers out there (like exari - it's been a while but I remember their stuff being pretty cool.)

That's a bit different from vanilla PDF generation, but perhaps a good portion of that is down to marketing. How you position this could be important. You've got your salesforce version - nice :-) That should mean you're selling direct to a business user, so have you worked out what their pain point is? Do they think of it as "generating PDFs" (you've branded it as "Great Reports", so obviously you think they don't!) I don't know anything about report generation on SF, but I'd be shocked if there wasn't already some competition - have you looked in to that? What's your point of difference?

Down to your assumptions: 1) I've only ever used FOP & yes it is tedious, but it gives you a lot of power, and that power is going to be hard to replicate.

2) Generating large no of PDFs is hard to scale. For a cloud service, though, you need to work out how you're going to deliver a large number of PDFs over the wire. (You'll easily be able to generate more bytes of PDF per $ of CPU than you'll be able to transfer per $ of bandwidth.)

3) That sounds like a safe assumption (almost a tautology).

4) I'm not so sold on this. If I can get equivalent functionality from a library, I'll use the library. I don't want to add a dependency on an external service (making my own service more fragile) unless it offers a significant benefit. Stripe saves me SO MUCH TROUBLE it's a no brainer. There's is no way I would even think of trying to replicate what they've got. The barriers are significant. What are the barriers to me using a library instead of Pagify? Not nearly as great.

Gut feeling, the SF report generation product is much more compelling, but check out the competition in that space.

knkella 13 years ago | |

Thanks for the feedback. We are going after the SF market, but there is already competition. As far as pricing model is concerned we would do that on the no of pages one generates. Do you think developers are not the right market? Can we offer it to developers in some form?

danielstudds 13 years ago | | |

PDF generation is a complex business... you're not going to be all things to all people, you need to choose a particular niche (at least to start with) and make that work. I can see the value prop in report generation on SF. "PDF Generation for developers" is too broad: I don't think you can do it (because there are so many different types of of PDFs to generate) and I can't see the value prop (because it's going to cost you more to provide a generic PDF generation service than it will cost me to use an existing library.)

That said, there might be a market there for developers, but you'll need to identify a very particular challenge that devs have and provide a much better solution than what they can achieve with a library in a day.

dbond 13 years ago |

It is tedious to layout a PDF but actual pain points are handling variable sized images and amounts of text, try formatting recipes to a PDF... Your demo video explains the features but not how it can solve anything that is a major problem to me.

The concept here in designing a template to be filled later is a brilliant idea, but I think its the delivery of the product that may pose the biggest problem. PDF generation doesn't have the same level of complexity as video encoding or payments so when it is used, its expected to just work as the environment doesn't change.

This introduces what I think is the main issue, if I build a website for a client, as a developer I never want to see it come back with problems, so its not worth introducing the risk of a third party service for the sake of a few hours of tedious work. At the other end of the industry any company producing a webapp may have problems with this simply due to how fundamental the feature can be, PDF generation doesn't require the same infrastructure as video encoding or payments, so they will have very different tradeoffs. Your main problem here is that you have no competition so I have nowhere to go if you close and the design online feature is essentially a vendor lockin.

Your assumptions are right but they are what a developer wants, the decisions are mostly made by management where future risk is more important than short term convenience.

Personally I think this product in particular would be much more suited to being a standalone package.

knkella 13 years ago | |

Thanks. Do you think by using an HTML template that can be filled in later is a better idea than visual designer?

dbond 13 years ago | | |

The visual designer is great, it adds back in the ability for the designer to alter the template without the developer having to change any code. If this was a standalone product the template format would most likely need to be a proprietary format anyway for some basic piracy protection.

LarryMade2 13 years ago |

Ive used FPDF/FPDI most of the time I used it to generate preformatted mailing labels (the best output method that is compatible with any printer, just format it with scaling "off" in mind.) Maybe a letter or report or two...

But the main thing that drew me to PDF utilities was to programmatically fill out pre-existing PDF forms. With FPDI i would load in the form and then overlay the data on top of it. Not as easy as charts and tables, as some data exists as one line per row and other might have several lines per row of data. different positions, etc. (here's a fun form I had worked with: http://www.cdss.ca.gov/cdssweb/entres/forms/English/CD9600.P... depending of family size or care situation could be multiple pages...) If you can crack stuff like that in an easy-ish to use API, I think you would catch more developers interest.

subsection1h 13 years ago |

    It is tedious to code the layout and design of the
    PDF using code libraries such as FPDF, Prawn etc.

I recently used Prince[1] to generate a PDF version of a web-based book. The PDF needed to be suitable for professional printing (crop marks, bleed, trim, etc.). Prince was the only solution I could find that came even close to being adequate for the job, and it exceeded my expectations. Creating the HTML for Prince wasn't tedious at all. It was almost a pleasure because Prince includes better support for print-related CSS3 properties than any rendering engine I've encountered. There's a web-based service named DocRaptor[2] that is powered by Prince.

[1] http://www.princexml.com/

[2] http://docraptor.com/

usladha 13 years ago | |

Hi,

You can try using FlyingSaucer (https://code.google.com/p/flying-saucer/) It takes in HTML and gives PDF. It also supports print related CSS3 properties and its open source. May be it cannot compete with Prince, but it is a good open source alternative

ashokvarma2 13 years ago | |

PrinceXML doesn't support a lot of the Html5 and CSS3. Canvas support doesnt exist, so I cant print charts.

usladha 13 years ago | | |

I havent tried charts with princexml, but flying-saucer has capability to do custom rendering. So if you want, you can use a service to get image of your chart and then render that in the pdf as image.

jyu 13 years ago |

The target market that will pay for PDF generation is probably not developers.

A competent developer can generate PDF's with the mentioned libraries rather quickly. We deal with a lot of PDF generation where I work, and it is not something that enters our minds to outsource because it's pretty fast and easy to do. For us, the generated PDF's do not need great design or layout, they just need the info in a printable format.

That said, you may find a better audience with less tech savvy users that have a lot of reporting needs. I can't really think of a target audience off the top of my head. Good luck to you.

ashokvarma2 13 years ago | |

I disagree, If we are dealing with simple invoices and stuff. Once you start dealing with reports containing charts, it all goes haywire

ashokvarma2 13 years ago |

Its a terribly painful. I have tried every single one of the options. I am still finding it extremely difficult to get it working. Once I managed to get it to print to pdf, then started the bigger problem 'Paging'.

Paging is quite painful. I am writing my own code to split tables to fit into pages.

If you can solve this problem. I would gladly pay.

knkella 13 years ago | |

We are looking into dynamic layout where tables can be split. It would be great to work along side you. Can I have your contact?

ashokvarma2 13 years ago | | |

Oh man, if you can solve it for me, I will be eternally grateful. You can email me at ashok (at) reportgarden (dot) com

palidanx 13 years ago |

I rolled up my own solution for rails here.

http://ror-snippets.tumblr.com/post/40142053834/converting-h...

Imo only a subset of people have pdf generation needs (mainly devs)

edoceo 13 years ago |

It's super easy here, I'm generating nice multi-page PDFs with colours and images and all that - from HTML5 based content. I have a similar API available at http://edoceo.io#pdf

jdietrich 13 years ago |

Not amongst the kind of people who read HN. You might get some traction with semi-skilled PHP developers, but they rarely have much of a budget to work with and will generally use whatever is the cheapest acceptable solution.

gee_totes 13 years ago | |

This comment is wrong. PHP is a total mess when it comes to pdf generation, even skilled developers have to resort to things like TCPDF.

This service would provide a good alternative, and regardless of what the developer is making, the price of the service would be far lower than paying the dev for the amount of time they spent futzing around with PDFs in PHP.

gee_totes 13 years ago |

Can I send a URL to the API and it will return a PDF? That is something I would use for sure.

knkella 13 years ago | |

Yes it will. Can we talk more about it?

gee_totes 13 years ago | | |

So I could send it www.google.com and it would return a .pdf of google.com, exactly as it would look in the browser?

factorialboy 13 years ago |

Browsers let you print a page as PDF. Which use-cases does Pagify solve?

knkella 13 years ago | |

It does print the web page. But when it comes to print-friendly reports and quotes you need some library or print ready web page. So we allow developers to create a template that is print ready. And then using an API call they can merge data and generate PDF.

knkella 13 years ago |

Help us understand PDF generation and its problems.