Cymen's Blog

At work we use Jenkins (formerly Hudson) for continuous integration and for pushing out releases to our staging and production servers. It works well for this however the configuration for msdeploy is a bit of a cluster due to it being in a batch script — adding or removing directories to skip for deployment is a pain.

I realized today one way to solve this might be to support creating a .skip or .skip-TARGET (so in our case, .skip-production and .skip-staging) files. Than add a batch or simple program to scan the project directory (for script, can use “dir /S /B FILENAME” seems to be equivalent of UNIX “find FILENAME”) and add a skip for the directory. That way anyone can add the skip option and it is clear in source control what is and what is not getting pushed out.

It seems like a simple enough idea to implement…

So there were a few issues with the previous version — at least when I wanted to extend it for some custom calculations so here is yet another version:

        public static int PercentComplete(this ModelStateDictionary modelStateDictionary, int? ScalePercentWithValueAsZero = null, int? MaxValue = 100)
        {
            int totalItems = 0;
            int validItems = 0;
            int percentComplete = 0;

            if (MaxValue.HasValue && (MaxValue.Value < 0 || MaxValue.Value > 100))
                throw new ArgumentOutOfRangeException("MaxValue must between 0 and 100!");

            if (modelStateDictionary.IsValid)
            {
                percentComplete = MaxValue.Value;
            }
            else
            {
                foreach (var item in modelStateDictionary)
                {
                    totalItems++;
                    if (item.Value.Errors.Count == 0)
                        validItems++;
                }

                if (totalItems > 0)
                    percentComplete = (100 * validItems) / totalItems;

                if (ScalePercentWithValueAsZero.HasValue)
                {
                    if (ScalePercentWithValueAsZero.Value >= percentComplete)
                    {
                        percentComplete = 0;
                    }
                    else
                    {
                        percentComplete = Convert.ToInt32(Math.Ceiling((double)(percentComplete - ScalePercentWithValueAsZero.Value) / (100 - ScalePercentWithValueAsZero.Value) * 100));
                    }
                }

                if (MaxValue.HasValue)
                {
                    percentComplete = percentComplete * MaxValue.Value / 100;
                }
            }

            return percentComplete;
        }

And if a model has a particularly complicated percentage complete calculation in which one needs to manually check some things and add to the total that can be done:

        public int CustomPercentComplete(Func<int?, int?, int> PercentComplete)
        {
            // scale percent complete 50-100% as 0-100%
            int basePercentAsZero = 50;

            // actually, scale 50-100% as 0-91%
            int max = 91;            

            // call default PercentComplete
            int percentComplete = PercentComplete(basePercentAsZero, max);

            // implement here your custom percente complete for the remaining 9%

            return percentComplete;
        }

An example of calling the custom percentage complete calculator on your model:

        myModelInstance.CustomPercentComplete(ModelState.PercentComplete);

And if you need to call either calculator but your controller action doesn’t bind to an instance of the model you need to calculate the percentage on, you can add a method in your controller like this one (hopefully there is a better way — let me know if you know of one):

        // Work around for getting percentage complete when in another action where the model is not the application
        // like so:
        // int percentComplete = MyPercentComplete(application);
        private int MyPercentComplete(MyApplication application)
        {
            return Convert.ToInt32(MyPercentCompleteAction(application).Content);
        }

        // Work around for getting percentage complete when in another action where the model is not the application
        // like so:
        // int percentComplete = Convert.ToInt32(MyPercentComplete(application).Content);
        private ContentResult MyPercentCompleteAction(MyApplication application)
        {
            return Content(application.CustomPercentComplete(ModelState.PercentComplete).ToString());
        }

Update: The mathematics is wrong — not sure what I was thinking. See the updated version here.

So the prior version has a bug — it works fine if the scaling value is 50 however it fails at other values. Whoops! That has been fixed and now we have a MaxValue scaler too. The MaxValue scaler is useful if you want to calculate the percentage done as maxing out at say 91% because you want to manually calculate the remaining 9%.

    public static class ModelStatePercentCompleteCalculator
    {
        // Very rough but reusable % complete calculator that is an extension so can be
        // called simply as ModelState.PercentComplete() in a controller action. Iterates
        // over the items in ModelState returns percentage complete where a complete field
        // is seen as having no errors. If the ModelState.IsValid returns true, the
        // model is 100% complete.
        //
        // ScalePercentWithValueAsZero: if you calculate the % done on a model and it says say 40%
        // really it is 0% than set ScalePercentWithValueAsZero=40 and it'll treat 40% as 0% and
        // scale it appropriately to 100 or maxValue
        //
        // MaxValue: if you want the maximum value to be less than 100 than set it here so you can
        // accomodate calculating to 100% based on some other criteria too (which you're
        // responsible for)
        public static int PercentComplete(this ModelStateDictionary modelStateDictionary, int? ScalePercentWithValueAsZero = null, int? MaxValue = 100)
        {
            int totalItems = 0;
            int validItems = 0;
            int percentComplete = 0;

            if (MaxValue.HasValue && (MaxValue.Value < 0 || MaxValue.Value > 100))
                throw new ArgumentOutOfRangeException("MaxValue must between 0 and 100!");

            if (modelStateDictionary.IsValid)
            {
                percentComplete = MaxValue.Value;
            }
            else
            {
                foreach (var item in modelStateDictionary)
                {
                    totalItems++;
                    if (item.Value.Errors.Count == 0)
                        validItems++;
                }

                if (totalItems > 0)
                    percentComplete = (MaxValue.Value * validItems) / totalItems;
            }

            if (ScalePercentWithValueAsZero.HasValue && ScalePercentWithValueAsZero.Value > 0 && percentComplete >= ScalePercentWithValueAsZero)
            {
                percentComplete = (percentComplete - ScalePercentWithValueAsZero.Value) / (MaxValue.Value - ScalePercentWithValueAsZero.Value) * 100;
            }
            else
            {
                percentComplete = percentComplete / 100 * MaxValue.Value;
            }

            return percentComplete;
        }

AutoMapper is great but it is both specialized, heavy duty and lacking in documentation. The lack of documentation can bite you: if you need to map two objects in different ways in different places things may not work as you expect. For example, say I have Mapper.CreateMap() with a number of .ForMember(x => x.Something, opt => opt.Ignore()). Now if I create the same mapping elsewhere but with different .ForMember options the prior .ForMember mapping options persist into my current map. To fix this, one can do a Mapper.Reset() which will blow away the global mappings. But creating mappings is expensive. Food for thought.

Update: there is a bug with this code. Please see this new post with updated code.

This version adds a baseline value so the percentage is scaled from that baseline (as 0) to 100%. This is to accommodate the case where the form starts out at say 50% complete. In this case, the call in the controller would be:

    x.PercentComplete = ModelState.PercentComplete(50);

Note this will cause some odd results if you pick a base value that isn’t truly a base value. In that case, the percentage calculation as the model is “completed” will increase, dip to zero at the false base value and then increase to 100 from there.

    public static class ModelStatePercentCompleteCalculator
    {
        // Very rough but reusable % complete calculator that is an extension so can be
        // called simply as ModelState.PercentComplete() in a controller action. Iterates
        // over the items in ModelState returns percentage complete where a complete field
        // is seen as having no errors. If the ModelState.IsValid returns true, the
        // model is 100% complete.
        public static int PercentComplete(this ModelStateDictionary modelStateDictionary, int? ScalePercentWithValueAsZero = null)
        {
            int totalItems = 0;
            int validItems = 0;
            int percentComplete = 0;

            if (modelStateDictionary.IsValid)
            {
                percentComplete = 100;
            }
            else
            {
                foreach (var item in modelStateDictionary)
                {
                    totalItems++;
                    if (item.Value.Errors.Count == 0)
                        validItems++;
                }

                if (totalItems > 0)
                    percentComplete = (100 * validItems) / totalItems;
            }

            if (ScalePercentWithValueAsZero.HasValue && ScalePercentWithValueAsZero.Value > 0 && percentComplete >= ScalePercentWithValueAsZero)
            {
                percentComplete = 100 / ScalePercentWithValueAsZero.Value * (percentComplete - ScalePercentWithValueAsZero.Value);
            }

            return percentComplete;
        }
    }

Did you know that AutoMapper’s Mapper.CreateMap() persists even if you use different options on the mapping? So if you have this in one place in your code:

Mapper.CreateMap()
    .ForMember(x => x.BrainSize, opt => opt.Ignore());
var person = Mapper.Map(anotherPerson);

And then later this:

Mapper.CreateMap();
anotherPerson.BrainSize = 5000;
var einstein = Mapper.Map(anotherPerson);

That second usage won’t map BrainSize! This is quite confusing… The cache makes sense but it would be more intuitive if the cache was sensitive to .ForMember() usage and other options so that the exact same mapping is cached not any mapping.

The solution is to use this (potentially both before and after your use of AutoMapper):

Mapper.Reset();  // don't want some other area's usage polluting us
// create some mappings here that we don't want to pollute other areas
...
Mapper.Reset();

I haven’t read the source yet but this is a very non-intuitive aspect of AutoMapper. Basically on the level of a deal breaker almost in my opinion.

Say you have a number of models that are mostly comprised of plain old CLR objects and you need a rough percent complete calculator. Assuming you are using the standard ASP.NET MVC validation rules (via attributes), here is a rough helper method that can calculate the percentage complete for the model based on the number of fields with validation errors divided by the total number of fields.

        public static int PercentComplete(this ModelStateDictionary modelStateDictionary)
        {
            int totalItems = 0;
            int validItems = 0;
            int percentComplete = 0;

            if (modelStateDictionary.IsValid)
            {
                percentComplete = 100;
            }
            else
            {
                foreach (var item in modelStateDictionary)
                {
                    totalItems++;
                    if (item.Value.Errors.Count == 0)
                        validItems++;
                }

                if (totalItems > 0)
                    percentComplete = (100 * validItems) / totalItems;
            }

            return percentComplete;
        }

Writing Custom Rules for Web Deployment Tool

This is some very useful information. It’s somewhat frustrating that msdeploy is similar to rsync but seems much less functional from the stand point of what is baked in for rules. And this approach is unofficial and all that so…

The best SQL formatter I’ve found so far is:
sqlformat.appspot.com
It works nicely with subqueries which is becoming increasingly important for my use of it.

Sometimes one needs to replicate one gem environment to another host. Maybe there is a ruby way to do this but here is a quick script that runs “gem list” and then makes a list of “gem install PACKAGE -v VERSION” commands. It’s quick and dirty but maybe it’ll be useful for you too:

#!/usr/bin/php -q
<?php

?>